Frontier and Society in

Roman North Africa

DAVID CHERRY

CLARENDON PRESS· OXFORD

I998

2 the loeation of the
The Roman Frontier-Zone
The same sort of thinking that interpreted the Roman conquest
of north Afnca as a bless1l1g for its indigenous peoples has
tradItlOnally understood the Roman frontiers to be lines of
demarcation that distinguished and separated the civilized (that
IS, Roman) world from the barbarism that was thought to lie
beyond it.' The jUdgement is probably also a consequence of the
nature of the physical record: very little that is visible survives of
the Roman fron tiers apart from their military installations-their
forts, roads, and linear barriers. So Hadrian's wall might be
understood to have been intended, as one ancient writer put it
(SHA Had. I I. 2), 'to separate barbarians and Romans' (qui
barbaros Romanosque divideret).2
It is a view that seems now to have been largely abandoned, for
at least two reasons: first, because none of the frontiers' linear
barriers appears to have been wholly defensive in purpose; sec-
ond, because the Romans themselves for the most part seem not
to have considered the fron tiers to be lines either of defence or of
demarcation. It has been remarked recently that the only surviving
1l1tact Roman map, the Tabula Peutingeriana, does not even identify
. I There is ,an excellent discussion of modern thinking about thc Roman frontiers
In c., R. Whlttaker, Frontiers o[ fhe Roman Empire: A Social and Economic Study
(Baltlmore, 1994), 1-9.
. z, S,ee e.g. S. L Dyson, The Creation o[ the Roman Frolltier (Princeton, 19 5),
3· 1t sep~rates thc world of Rome from that of thc barbarian'; R. MacMullen, 8 7 Whittaker: 'Trade and Frontiers of the Roman Empire', in p, Gar?sey and C.
,~(1a,:g~S ',n fhe Roman Empire: l!ssays in ,'lIe Ord~nary (Princetol1, 1990), 49: like
sllnilal lInes along other frontJers latcr, Hadnan's wall 'indicated a wish to
~eparate Roma? from non-Roman by a permanent barrier'. So, tao, A. R. Birley,
Rom~n . Fr?ntJers and, Roman Frontier POlicy: Same Refiections on Roman
Io:peuahsm, TransactlOl1s oJ the Architectural and Archaeological Association
oJ Dur~am a~d !'orthumberland: 3 (1974), J6-17. See also E. L. Wheeler, 'Meth-
O~:logleal LllUlts and the MIrage of Rom.an Strategy', Journal 0/ Military
border'. 57 (I993), 29, where the /ossatum m north Afriea is deseribed as a
[-!t,<,!ory,
The Roman Frontier-Zone 25
frontiers 3
And the Latin word that modems
routinely employ to describe the Roman frontier-limes-was in
Roman times a surveyor's term adapted in military usage to mea~
'road', and not widely used to denote a demarcated border unlIl
the fourth century AD 4 . . .
It seems that the Roman fron tiers have always been descnbed m
spatial terms. So where they were once 'boundaries', for example,
they are now generally understood to have been 'zones'. 5 I,t may be
agreed with C. R. Whittaker that the fronlIers functlOned necessa-
rily' as zones, in so far as the Roman government ,:as un~ble to
achieve, as Owen Lattimore has said also of the fronlIers of Imper-
ial China,6 the 'optimum balance between its range of conquest (1.e
its militaryeapacity) and the economy of its rule (i.e where the
military expenditure is no longer paid for by tax returns); a?d
because the turn-over from economic viability to eCOnOlTIlC hablhty
is necessarily gradual, unperceived and unstable'.7 But it eannot be
demonstrated that the Romans themselves ever understood the
frontiers to have behaved, militarily or administratively, as zones,
or for that matter as any other kind ofterritorially defined unit. The
imperial government's demonstrated desire to extend its influence
weil beyond the line of the fron tier fortifieations suggests that, as
Benjamin Isaac has recently put it, the eoncept of a fronlIer was of
littIe real importanee in shaping military strategy8
3 B. Isaac, The Limits 0/ Empire: The Roman Army in the East, rev. edn, (Oxford,
2 8
), 39B. . Isaac, 'The Meaning of "LImes
J994 See . " an d "L"Imltan~l'".In A nelen,
. t Sources' '
JRS 78 (1988), 125-47; Limits 0/ Empire, 408; see also Whlttaker, ,Front/ers, 20~,
5 Boundaries: e.g. 1. C. Hudson, 'Theory and Methodology ~n Comparatl.ve
Frontier Studies', in D. H. Miller and J. O. StefTen (eds.), The Fron.ller: Compara,flVe
Studies (Norman, Okla., 1977), 12-13. Zones.: see espec.ially WhlUaker, Frontwfs,
62; cf. H. Elton, Frontiers 01 the Roman Empire (Bloommgton, lnd., J996), 4, 113
('overlapping zones').
6 Studies in Frontier Histvry (London and New York, 1962), esp. at I J3 .
R, Whittaker (eds.), Trade (lnd Famine in Classical,Antiquity (Cambndge, 1983),
JJ3; cf. Frontiers, 85: the frontiers were 'a compromlse betw~en the range
quest and tbe economy of rule'; 'Supplying .the System: Frontle~s and Beyond , m
0: ~on­
!.
c. Barrett, A. P. Fitzpatrick, and L. Macmnes (eds.), Barbanans and Romans In
NOJ·th- West Europe: From the Loter Republic to Late Antiquity (Oxford, 1989), 65:
the frontier was a 'borderland'. " ..
8 Isaac: Limits o/, Empire. 409, 417; see also M. Euzennat, L OIlVler et le
"Limes" considerations Sllr Ja frontiere romainc deTripolitanie', BCTH (1985),
qr; 1. C. Mann, 'Power, Force and the Frontiers of the Empire', JRS 69 (J979),
175-83; Whittaker, Fron/las, 95.
 The Roman Frontier-Zone 27
It might be uncIear, even to the Romans, whether certain indi-
genous peoples were within the empire's frontiers: so) for example,
the highland tribes of the Middle Atlas mountains in Mauretania
Tingitana (mod. MorocCO)9 And because the frontiers were often,
it seems, little more than 'the frozen forward lines of advance that
could be held following military campaigns', wand therefore inher-
ently fluid and unstable (or, put another way, 'dynamic')," it is
unlikely that many Romans, incIuding even the soldiers on the
ground, would have recognized or understood any of the kinds
of demarcated borders that can be drawn on a map (in the way that
the fron tier-zone in Roman-era Algeria is represented in Fig. 2. I,
for example).
Perhaps we would do better to deline the Roman fron tiers in the
same way that historians of the western United States have long
described the American frontier, that is, as a cultural process. 12 The
fron tier in Roman north Africa may be understood to have func-
tioned, independently of space (though not of time), as a zone of

.~ 9 B. D. Shaw, 'Autonomy and Tribute: MOU11tain and Plain in Mauretania

Tingitana', ROMM 41-2 (1986), 86 n. 69; on the frontier in Mauretania Tingitana,
see M. Euzennat, Le Limes de Tingitane: La Fronfiere meridionale (Paris, J989).
10 Isaac, Limits o/Empire, 417; see also Mann, 'Frontiers', 513-14, who suggests

1ha1 the frontiers arose 'by default'.

11 On 'dynamic' (as opposed to 'static') frontiers, see especially 1. A. Alexander,

'Frontier Studies and the Earliest Farmers in Europe', in D. Green, C. Haselgrove,

and M. Spriggs (eds.), Social Organisation and Settlement: Contributions fr0111

N
I-
""
0
Anthropology, Archaeology, and Geography (Oxford, 1978), 13-29; 'Early Frontiers
in Southem Africa', in M. Hall ef al. (eds.), Frontiers: Southern African Archaeology
.ß Today (Oxford, 1984), 12-23; 'The End ofthe Moving Frontier in the Neolithie of
North-Eastern Afriea', in L. Krzyzaniak and M. Kubosiewicz (eds.), Origin and
"
0
Early Development 0/ Food-Producing Cultures in North-Eastern A/rica (Poznan,

,
I
I
"'
~
0
~
1984), 57-63; cf. 1. H. F. Bloemers, 'Aeculturation in the Rhine/Meuse Basin in the
Roman Period: Demographie Considerations', in Barrett, Fitzpatrick, and
-,
I ] "
0
"
I.
Maeinnes (eds.), Barbarians and Romans, 178, and, on the Ameriean frontier,
I
!J .~ M. Mikesell, 'Comparative Studies in Frontier History', in R. Hofstadter and
"
"
I 8 M. Lipset (eds.), Turner and the Sociology of the Frontier (New York, 1968), 154.
I 'x 0
W. 1. H. Willems, 'Romans and Batavians; Regional Developments at the Imperial
'1
\. i"- ... 2 ~
,, ~ Frontier', in R. Brandt and 1. Slofstra (eds.), Roman and Native in the LOIv

\
"'"
-<C &:; Countries: Spheres of Interaction (Oxford, 1983), 106, would have us distinguish
instead between 'imperial' (i.e. dynamie) and 'eolonial' (i.e. statie) frontiers. See
~
I ..;
'. N
also D. 1. Mattingly, Tripolitania (Ann Arbor, 1994),77, who describes the army's
,
'1
?:2 ob
advance in north Afriea as a 'rolling frontier' .
12 See e.g. R. F. Berkhofer, Jr., 'The North Ameriean Frontier as Proeess and
\
\
"--.::: [;
Context', in H. Lamar and L. Thompson (eds.), The Frontier in History: Nm·th
America and SoutheJ'n Africa Compared (New Haven and London, 1981), 43;
w. P. Webb, The Great Plains (New York, 1931), on which see Whittaker,
Frontiers, 5.
 28
The Roman Frontier-Zone
interpenetra.tion between two previously distinct societies. 13 I am
not suggestmg that the development of the frontier-zones was
unaffected by their physieal or environmental charaeteristics.
LocatlOn wIll have determined, at the very least, whieh peoples
were admItted to the fronlIer-zones. My position rather is that the
fronlIers cannot. be shown to have perfonned any historically
recoverable functlOn other than to have aecommodated the contaet
of Roman and indigenous society.
. The real question then (and the issue that much of this book is
mtended to explore) is whether the frontier-zones served, as
Whlttaker has suggested, to unite and integrate those who were
'culturalI~ diverse',. or, p~t another way, whether they were, as one
comparatlve [rontIer l11storian believes, 'fron tiers of inclusion
(more properly assimilation)'. '4 lt will be argued that in the case
of the fron tier-zone in Roman-era Algeria, all the a~ailable evi-
dence mdICates that there was very little in the way of eultural
mtegratJon.
. My purpose in this chapter is to describe the nature and func-
tlOn of the Algenan fron tier: its forts and fortifications, roads, and
Imear barrrers (the jossatum). My position, which is elaborated
below, IS that the fron tier cannot be shown ever to have had a
stnetly defensIve purpose. lts development and arrangement
would seem to mdlcate ll1stead that it was intended primarily to
provlde for the secunty of the soldiers who were stationed in the
regIOn, and to enable the army (and, indirectly, the imperial gov-
ernment) to tax the products of pastoralism.
LOCATING THE ROMAN FRONTIERS
Modern thinking about the factors that determined the location of
the Roman frontiers seems often to be bound up with ideas about
3 nisse der Forschung über die Grenzprovinzen des römischen Reiches: Vorträge beim
,1 Cf..W S. Hansol~, The Nature <:lnd FUl1ction ofRomal1 Frontiers' in Barrett
~ltzPbatnck, aud Macmne~ (eds.), Barbarians and Romans, 55 ('an area' of interac~
t100 etween two cUltures). On 110rth Africa see C. M WeHs 'Th p. bl f '7 cr Mattingly, Tripolitunia, lO7, on thelinear barriers (cfausurae) ofTripolitania.
Desert Fro t" , . V A M ' ., e 10 ems 0
'. n lers, m . . axfield and M. 1. Dobson (eds.) Roman F, {.
Stud~r:s 1989: Proceedings C!l
Studles (Exetcr, I991), 47 8.
theXVth International Congress 0/'
Roman F;~~{:;;
'4 W~ittaker:, Frontiers, ,72-3; c~ 'Trade and Frontiers', 12I. 'Frontiers of inc1u-
sion': MlkeselI, Comparahve StudJCs', 153-4 (his itaJics). See also Elton, Frontiers,
10 9.
The Roman Frontier-Zone 29
their purpose. It has been said recent1y, for example, that 'the
factors whieh influenced the precise chaiee of fron tier line after
advanee or retreat are directly related to the function it was
intended to perform'. '5 It used to be widely thought that the
loeation of the fron tiers was determined mainly by military, and
especially by defensive, considerations. So it might be said of north
Africa's complex array of fortifications and linear barriers that it
was intended to protect roads and lines of communications against
incursions of Saharan nomads. 16
There is, in fact, little reason to think that the Romans' purpose
in north Afriea was ta proteet the region from those who lived
beyond the frontier. If the jossatum, far example, was really
intended to defend what lay bebind it (against amenace real or
imagined), which is to say that it was meant to interdict, in some
measure, access to or across the frontier-zone, we might reasonably
expect it to have been situated at or ne ar the zone's outer limits.
How are we to explain then that its component parts are, for the
most part, somewhere in the middle of what is generally agreed to
have been the Algerian fron tier-zone (as Figs. 2. land 2-4 indi-
eate)?" Either we must assurne that the Roman authorities charged
with deeiding its location eleeted to defend that part ofthe frontier-
zone which was located behind, that is, north and east of, the
[ossatum, but not, for some reason, the part that lay beyond, tbat
is, to the south and west of, it (what is represented as 'B' in Fig. 2.2),
ar it must be admitted that it had 110 real defensive purpose at all.
There are other reasons for thinking that the frontier in north
Africa was not meant to be a barrier against 'tribai hostility'. IX The
/5 Hanson, 'Nature and Function', 58.
16 H.-G. Pflaum, 'La Romanisation de l'Afrique', Vestigia, I7 (r973), 63. L.
Leschi, Etudes d'epigraphir:, d'archeologie ef d'histoire africaines (Paris, 1957), 32,
is similar. See also P. MacKendrick, The North African Stones Speak (Chapel Hili,
NC, 1980), 247= thejossatum was designed to 'control or keep out nomads'. Cf. E.
Birley, 'Hadrianic Frontier Policy', in E. Swoboda ef al. (ecls.), Carnuntina: Ergeb-
internationalen Kongress der Altertums/orscher Carnuntum 1955 (Graz and Cologne,
1956),29 (on Hadrian's wall).
/8 'Tribal hostility': E. L. Manton, Roman North Africa (London, 1988), 9I. The
notion is rejected by Mattingly, Tripolitania, 79, 107; R. Rebuffat, 'Au-deli des
camps romains d' Afrique mineure: Renseignement, contröle, pem~tration', ANRW 2/
loh (1982), 508; see also W. S. Cooter, 'Preindustrial Frol1tiers and Interaction
Spheres: Prolegomenon to a Study of Roman Frontier Regions', in Miller aod
Steffen (eds.), The Fronlier: Comparative Studies, 84.
 3° The Roman Frontier-Zone
N seems to have been about 6-9 feet across at the surfaee, 3 feet
Prontier-zone
t argued below that it was meant to interdict, not people, but
Frontier-zone
Desert
Fig. 2.2. Linear barriers andfrontier-zones
very nature of the /ossatum, for example, makes it unlikely that it
was ~eslgned :0 prevent ?f even to deter anyone from crossing the
fr?lltIer-.zone. 9 T~e ~arner was discontinuous, with gaps up to 40
mIles wlde, and m Ils most thoroughly studied sector known
locally as the 'Seguia Bent el Krass', which stretched ~bout 37
mIles east-west across the poor-quality winter grazing-land that
lIes south of the Oued Djedi, it consisted of little more than a low
mud-bnck wall, wlth an apparently irregular system of widely
( 19 ~t cannat be said oftheJossatum, however, as it has been said ofHadrian's wall
y E. N. Luttwak, The Grand Strategy 0/ fhe Roman Empire jrom the Fi t
e.g.
~en~urY.A. I? ta. fhe Third (~altimore, 1976), 66), that the planting of forts beyo~~
e arner IS eVldence that 1t was not meant s'olely to serve a defensive purpose for
n~ne ?f. the forts located south-west of it Ce.g. EI-Gara, Ai'n Rich Caste{lum
Dlml?Jdl) cao. be. show to hav~ been built belore the lossatum (which is almost
c~rt~m~ H~dnall1c). The e~tenslOn of a military presence to the region south(~west)
o t e arner tells us nothmg about why it was constructed in the first place.
The Roman Frontier-Zone 31
spaced gateways and towers, fron ted by the ditch itself, which
wide at its base, and perhaps 6-9 feet deep.oo It will be
animals.
Th,re is no basis either for assuming that Roman offieials,
including the emperor, always eonsidered it to be their duty to
protect provincial populations against the peoples who lived
beyond the fron tiers" (though they may sometimes have been
expeeted to do so, perhaps especially from about the middle of
the third eentury AD). And even if the imperial government (or the
emperor) had wanted to implement a poliey that aimed at devel-
oping a coherent system of defensive barriers and fortifications, it
is doubtful that it could have overeome what Fergus Millar has
called 'the fundamental limitations' whieh were 'placed by time,
space and delays of communication', and which were a necessary
consequence of the vast distanees that separated the frontiers from
the capital (and from each other)." Isaac is undoubtedly right in
thinking that the location of the frontiers was rarely, if ever,
determined by a conscious deeision to establish a fixed line of
defensive positions. 23
A variation on the view that the frontiers were meant to be
preclusive barriers has their location determined by 'geography'.24
One recent study of the western fron tiers of the empire concludes
20 The most complete description of the Seguia Bent el Krass is in 1. Baradez,
Fossatum Africae: Recherehes aeriennes sur l' organisation des confins sahariens ci
l'epoque romaine (Paris, I949), 93-I08; see also C. Daniels, 'The Frontiers: Africa',
in 1. S. Wacher (ed.), The Roman Warld (London and New York, I987), i. 244;
Leschi, Etudes d'epigraphie, 36-8. For the dimensions of the ditch, Y. Le Bohec, La
Troisieme Legion Auguste (Paris, 1989), 370 n. 34 (illustration at 429); cf. E. W B.
Fentress, Numidia and {he Roman Army: Socia!, Military and Economic Aspects 0/
the Frontier Zone (Oxford, 1979), 112 (2-4.5 m. wide). On the gaps, see also Whit-
taker, 'Trade and Frontiers', 122-3 n. 10.
21 The point is made in Isaac, Limits 0/ Empire, at 393-4 (cf. 377, 425). On the
role ofthe emperor in detennining fron tier 'policy', see especially F. Millar, 'Emper-
ars, Frontiers and Foreign Relations, 3I B.C.-A.D. 378', Britannia, 13 (1982), 7, who
argues that most tactical and strategie decisions were 'taken by the Emperor in
person'. 22 Miliar: 'Emperors, Frontiers', 7.
23 Isaac: Limits ofEmpire, 387; cf. 416-I7 ('military policy was dictated by events
in the field'). It seems to me to be pressing the point too hard, however, to argue that
'random factors often decided the objects of Roman wars' (379; my italics).
24 e.g. D. Potter, 'Empty Areas and Roman Frontier Policy', AJP r 13 (I992), 26 9
(citing also 'the cultural development of border regions' and 'military necessity').
 32 The Roman Frontier-Zone
that the imperial government 'searched' for defensible borders' the
legions, it is said, were moved to 'firm geographical feature~ on
which to fix their lines'.'5 But there is nothing to indicate that
Roman military planning was determined or even influeneed by
regional or loeal geographie conditions. The Romans cannot be
shown to have used maps in strategie or tactieal planning. ,6 And
the irregularity of the line of Roman fortifieations in many parts of
the empire, where it sometimes ignores cr even cuts across
eonspicuous geographie features, strongly suggests that they either
knew or (more likely in my view) eared little about geography. '7
Other historians, like lohn Eadie, have argued that 'relations
with tribal groups determined the shape and eharaeter of Roman
[rontiers'.28 The frol1tier-lines, it is said, were located, as a matter
of policy, at 'pre-existing triba! interfaces or lines of articulation'
in other words, in the border regions that separated one triball;
eontrolled territory /fom another. '9 At a broader level, the objeet
of Roman imperialism is understood to have been, in Isaac's
words, 'ethnic father than territorial or geographie'. 30
In a somewhat more sophistieated version ofthe same argument,
the limits of Roman expansion, and therefore, by default, the loca-
tion of the frontier-lines, are thought to have eoineided roughly
with the outer perimeter ofthe provineiallands that were eontrolled
by indigenous tribes whose soeial strueture was readily adaptable to
the Roman administrative system, whieh in turn is understood to
have aimed at governing the provinces indireetly through existing,
urbanized (or partly urbanized), indigenous social elites. The fron-
tiers, in other words, were meant to incorporate those triba! terri-
tories in which 'the adoption of native aristocracies was clearly
more straightforward and eomprehensive' beeause of their 'pre-
vious tradition of urbanism and developed central authority'.3 1
25 S. K. Drummond and L. H. Nelson, The Western Frontiers o/Imperial Rome
(Armonk and London, 1994), 6-7 ('The Search for Borders').
2{) See especially Miliar, 'Emperors, Frontiers', 18.
27 Cf. Isaac, Limits 0/ Empire, 401.
28 'Civitates and Clients: Roman Frontier Policies in Pannonia and Mauretania
Tingitana', in Miller and Steffen (eds.), The Fromier: Comparative Studies 59.
29 Cootel; 'Preindustrial Frontiers', 85. ' Northern Europe from AD 1-400', in M. Rowlands, M. Larsen, and K. Knstlansen
3.0 .Limits 0/ Empire, 395: cf. 426: 'Roman conceptions of power and military
actlVlty focused on peoples and towns rather than geography.'
. 3' ,~. Millett, 'Romanization: I:Iistorical Issues and Archaeological Interpreta~
tlOns, 111 T. F. C. Blagg and M. MIllett (eds.), The Early Roman Empire in the West
(Oxford, 1990), 38-9.
The Roman Frontier-Zone 33
Roman military expansion in Germany, for example, is interpreted
as having stopped at the interseetion of 'the more highly stratified
Celtic soeiety and the less stratified Germanie soeiety', beeause the
former, with its 'centralised political system' was 'far easier for the
Roman state to incorporate'. 32
It is undoubtedly true that the Roman authorities preferred to
govern the provinces indireetly, and did so wherever it was practi~­
able. It cannot be disputed either that the imperial government WIll
have wanted to deal with provincial societies that were eontrolled
bya native aristocraey whose members eould be readily eo-opted to
share in the benefits, and burdens, of Roman rule. What eannot be
demonstrated, however, is that Roman military or politieal objec-
tives, and therefore, in some measure, the loeation of the frontier-
lines, were determined on the basis of what was known about
indigenous soeieties. The implicit assumption, that the Romans
had suffieient knowledge of the peoples who lived near ar beyond
the frontiers, at the time that theji-ontier-lines were established, to be
able to distinguish between those who were adaptable to Roman
administration and those who were not, is almost certainly wrong.
There is nothing to indieate that the state systematieally eolleeted
information about the lands ar peoples who lay beyond the fron-
tier-zones; it may be agreed with Millar that the fron tier often
functioned as an 'information barrier'.33 Nor can it be shown
that the fron tiers traeked the interseetions of tribal eultures. In
fact the German frontier-line cut across a zone of social and
cult~ral homogeneity, dividing, as Whittaker has remarked, the
peoples who were neither wholly Celtie nor German 34 .
My own position is that the main purpose of the frontJers and
the factars that determined their location were, in a broad sense,
strategie and eeonomie. 35 It seems now to be widely agreed that the
frontiers' linear barriers were intended, not to prevent movement
into 01' across the frontier-zones, but to 'control' it. 36 It has been
said recently of Hadrian's wall and the German palisade, for
2
3 L. Hedeager, 'Empire, Frontier and the Barbarian Hinterland: R<?m.e and
(eds.), Centre and Periphery in the Ancient World (Camb~id.ge, 1987), .126.
33 MilIar: 'Emperors, Frontiers', 19. So, tao, Isaac, Limits 0/ Empire, 402.
34 Whittaker: 'Trade and Frontiers', III; 'Supplying the System', 66.
35 Cf. Whittaker, 'Trade and Frontiers', II3 ('economic, ecological and social') .
6
3 See e.g. D. J. Breeze and B. Dobson, Hadrian's Wall (Landon, T976), 77;
Mattingly, Tripolilania, I 14-T 5.
 34 The Roman Frontier-Zone
example, that 'the principle which they have in common' is the
'restrict~o~ a.~d contral cf movement into Roman territory', and
that thelr mIlItary function' was to 'facilitate the doser control of
small-scale local movement across the frontier'37 On one level at
least, this interpretation is, it seems to me, undoubtedly correc!. It is
in fact most conspicuously evidenced in the ca se ofthe north African
Jossatum, which, because it was discontinuous cannat have been
intended to prevent movement into Roman-con;rolled territory, but
must mstead have been designed to direet and ehannel it, in my view
not away [rom certain areas (for example, from 'the sown' as has
sometimes been argued),3 8 but rather towards the pla;es-the
leglOnary base at Lambaesis, for example-where it could be more
easily monitored by the Roman army, and, in some instances, taxed.
This is admittedly not a completely novel idea. It has been said
before that the fossatum was designed to 'keep an eye on the tribes
of the deser!', and to 'control and to direct the flow of traffie' into
and out of the frontier-zone. 39 The question that seems rarely to
have been asked, however, is this: why would the imperial govern-
ment have mvested so much of its resources (in money and in
manpower) in constructing and maintaining an elaborate system
of barriers and fortifications whose purpose was merely to monitor
(that is, 'watch over') the loeal, and in the north African case,
mostly seasonal movements of the trans-fron tier populations?40 It
has been suggested that the Roman linear barriers had a secondary
purpose of levymg customs duties on goods that were brought
across the fron tiers. 4' It will be argued below that this is very likely
to have heen the primary purpose of the fossatum.
3; Hanson, 'Nature and Function', 58-9.
3 e.g. ~Y Ba:ad~z, Foss~tum Africae, '48; Leschi, Etudes d'epigraphie, 71; see
also the dlScusslon III Mattmgly'Tri')olitania
39", 1 - , pp. xv, 69 .
Tnbes of tl?e desert': S. Raven, Rome in Africa, 3rd edn. (London and New
York, 1993),91. Flow of trafik': Fentress Numidia II2
40 'Watch over': Le Bohec, La Troisierne' Legion A~gus;e 578' see also M Janon
'~echerc.hes.a Lambese', An!. afr. 7 (1973),198-9; p. Tr~usset, Recherch~s sur l~
~lmes Tn~~iI(an!,s d~ Cho~t el-Djeri~~ lafrontiere tuniso-libyenne (Paris, 1974), esp.
at 140 -1, SJgmficatlOn dune frontIere: Nomades et sedentaires dans la zone du
limes d' Afrique', in W S. Hanson and L. 1. F. Keppie (eds.), Roman Frontier Studies
1979, (Ox~ord" I?80), 935; 'L'Idee de frontiere au Sahara d'apres les donnees
archeologlques,}n P"R. Baduel (ed.), Enjeux sahariens: Table ronde du Centre de
Recherehes et d'Etudes sur fes Socil!tes Mediterraneennes (Paris 19 81 ) 62
4' H 'N d ., , ,.
a~son, at~re an Funchon, 60; for north Africa, see 1.-P. Darmon, 'Note
sur le tar.lf de Z~rat'.' Cahiers de Tunisie, 12 (1964), 7-23; R. Mariehai, Les Ostraca
de Eu Njem (Tnpoh, 1992), 1I2-I3; Mattingly, Tripolitania, II3-!4; R. Rebuffat,
The Roman Frontier-Zone 35
The 'least-effort subsistence model' of frontier-zone economlc
development that has been constructed by eomparative historians
of agricultural eolonization predicts that frontiers will be located
where 'full extension oflands is possible', with the smallest possible
investment of labour and eapital, and 'within the limitations of
natural environmental faetors' (for example, drainage, preeipita-
tion, vegetation cover)42 Put another way, the frontiers are likely
to be located in regions where there is an economic and ecological
transition [rom intensive agricultural production to more extensive
uses of the land, in many eases therefore to pastoralism (in the
widest sense of the term).
The determining factor then in the loeation of the Roman fron-
tiers will have been, as Whittaker has remarked, 'the marginality of
the land' 43 The point is most dearly illustrated by the historical
development of the Algerian fron tier.
THE DEVELOPMENT OF THE FRONTIER-ZONE
It was remarked earlier that the Roman conquest of north Africa
was aecomplished over a very long period of time, beginning with
the defeat of Carthage in the third Punic war (149- 146 BC), and
achieving its fullest expression, in a territorial sense, only with the
Severan-era extension of forts into the fringe of the Sahara (for
example, Castellum Dimmidi, constructed in AD 198)44 The slow
pace of Roman expansion may be explained, at least in part, by
what seems to have been the popular understanding of the char-
acter of the African interior, which as late as the time of Pompo-
nius Mela (e. AD 40?) was considered to be generally worthless. The
'Une zone militaire et sa vie economique: Le Limes de Tripolitaine', in Armees el
Jiscalite dans le monde antique (Paris, 1977), 405-7; 'La Frontiere romaine en
Afrique: Tripolitaine et Tingitane', Ktema, 4 (1979), 230-5.
4
2
S. W Green, 'The Agricultural Colonization of Temperate Forest Habitats: An
Ecological Mode!', in W W. Savage and S. I. Thompson (eds.), The Frontier:
Comparafive Studies, ii (Norman, Okla., 1979), esp. at 78.
43 Frontiers, 86; cf. 288-9 n. 36; see also P Trousset, 'Limes et frontiere clima-
tique', in Actes du Ille colloque international sur l'histoire ef l'archeologie de l'Afrique
du Nord (Montpelliel; 1-15 avri/, 1985) (Paris, 1986), esp. at 65. I am not persuaded
by Wheeler's suggestion ('Methodologieal Limits', 234) that the Romans would
have been unable to identify economic marginality without 'demographie and
economie surveys'.
44 For the date, AE (1948),214; see also Lesehi, ifudes d'epigraphie, 322.
 The Roman Frontier-Zone
Romans will have had little interest in acquiring control of lands
that were thought to be unsuitable for cultivation.
Mainly by building a complex network of roads and forts, and
by imposing treaties and diplomatie controls (backed up by the
exchange of envoys and by the threat of military intervention),
Rome sporadically extended its military authority south and west
from the territory that had been controlled by Carthage, settling
eventually along a line-in east-central Algeria and in southem
Tunisia-that more or less coincided with the ISO millimetre
isohyet. 45
Modern opinion has it that nearly all of the roads were
constructed in the period between Augustus and Septimius
46
Severus Some are dated securely by their milestones (of which
more than 2,000 survive). More often, the milestones serve to
indicate only when an emperor began to advertise his generosity
in having provided a road 47 For the most part, and always out-
side the towns, the roads were unpaved paths, levelled off and
c1eared of stones48 Some at least are likely to have followed exist-
ing (that is, indigenous) routes. But all were characteristically
Roman in their determination to take the shortest possible line,
even where it meant ascending a steeper slope than necessary. They
generally kept to higher elevations, avoiding valley-bottoms when-
ever possible; where they did cross lower ground, they were often
seeured by watch-towers or lookout postS. 49
The roads seem to have been intended to serve one main
purpose, to carry military traflk 50 It would appear that some,
like the road built under Augustus from Ammaedara (mod.
Hajdra) to Tacapae (mod. Gabes), may have been designed also
to 'mark off' the land, in some fashion-the boundary-stones
45 On diplomatie eontrols, see Mattingly, Tripolitania, 74. For Tunisia, M.
Euzennat, 'Quatre annees de reeherches sm la frontiere romaine en Tunisie mer-
idionale', CRAI (1972),7-27; 'Les Recherehes sur la frontiere romaine d'Afrique',
in 1. Fitz (ed.), Akten des Xl internationalen Limeskongresses, Szeke~fehervlll; '976
(Budapest, 1977), 533-43.
6
4 Ravell, Rome in Alrica, 66-7. On Roman roads in north Africa, see especially
P. Salama, Les Voies romaines de I'Ajriqu(' du Nord (Algiers, 1951).
47 Isaac, Limits 01 Empire, 34 n. 94.
4R R. G. Goodehild, 'The Roman Roads of Libya and their Milestones', in F. F.
Gadallah (ed.), Libya in History (Benghazi, 197I), 157.
49 See Manton, Roman North Alrica, 84-5.
5° Cf. Isaae, Limits 01 Empire, 103 (roads are 'better interpreted as links than as
barriers').
The Roman Frontier-Zone 37
associated with them (eIL 8. 22786, for example) employ the word
limitavit. 51 . .
Forts, too, were scattered widely across the regIOn, especmlly
over the southern part of what was to become the provlllce of
Numidia and in Mauretania Caesariensis. The factors and
consider~tions that will have decided their locations are, unfortu-
nately, mostly unrecoverable, at least in part because the Roman
sources never discuss them 5 ' Their location was probably deter-
mined partly by indigenous political geography, mcludmg the
distribution of proto-urban centres. 53 The legionary base at
Lambaesis, for example, seems to have been built at the site of a
pre-Roman urbanized settlement. 54
It might be assumed that at least some forts were placed at or
near strategically important sites such as tnbal centres and bound-
aries. 55 But the only factor that can be shown consistently to have
influenced the choice of sites is the availability of water: so a
number of forts in southern Numidia-at Ad Maiores, AYn RICh,
Doucen, Gemellae-were located near spnngs . or a t d eser t 0 ases. 56
The Frontier in the First Century AD
There were enough Roman citizens living at Cirta by 112 BC for
Jugurtha to be able to massacre a significant number of them 57
But it was not, it seems, until the time of Augustus that there were
5' That limito was used to mean 'mark off' (or s?m~thing like ~t) at least fro~
about the middle ofthe Ist cent. AD would seem to b~ 1l1dlcated by Plmy, HN 17· 169.
vineas limitari decumano XVIII pedum latitudinis ('v1l1eyards should be sepa~ated by
a roadway eighteen feet wide'). . 5~ ~ee Isaac, Lin-:i~S o.f ~mp,lre, 375.
53 Cf. M. Millett, The Romanization 0./ Bntam: An Essay 111 Afchaeologlut!
Inter relation (Cambridge, I990), 74; R. 1. A. Wilson, JRS 8,2 (1992), 29 0-3 (01:
Bn.('aln.
P ) 0 n Wales , see V.E. Nash-Williams, The Roman Frontzer . , mb Wales (Cardlff,
tl't
1954), who concludes (7) that there was a 'si~nificant eo~forI?-lty etween le SI es
of Roman forts and the principal coneentratlOns of natIve IulI-fort settlements.
54 Janon 'Recherehes a Lambese', 217. , .
55 Cf.M: Millett, 'Forts and the Origins of Towns: Cause or Effec.t? : 1ll T. F. C.
·
BI <lggan d A. C King
. . (eds) , Military and Civilian in Roman"1 Bntam: ' t 'Cultural
d
Relationships in a Frontier Province (Oxford, 1984), 69 ..A snll1 ar pom IS ~a. e
about Seotland in L. Keppie, 'Beyond the Northern FrontIer: Roman and Native III
Sco11an d ','n M Todd (ed.) Research on Roman Britain 1960-89 (Londoll, 1989),
I . , G f ·1' T Mur)
67. For strategically loeated forts in Wales (e.g. Bryn-y- e el laU, omen-y- ,
see Nash-Williams, Roman Frontier, 111. . . ,.
56 Cf. Fentress, Numidia, 72, who emphasizes instead proX111uty to eultlvab~e
land. Does she mean to imply that the soldiers were expected also to ?e far.mers.
57 Sallust, Bell. lug. 21 (multitudo togatorwn); see also Raven, Rome m Ajnca, 52.
 The Roman Frontier-Zone
any sustained Roman military operations outside the territory that
had been controlled by Carthage. Four governors of the province
of Africa Proconsularis were awarded triumphs in 34, 33, 28, and
21 BC: it is unclear where they fought, or against whom 58 In 19 BC,
L. Cornelius Balbus directed an expedition, apparently to the
Fezzan, against the Garamantes, for which he, too, was awarded
a triumph. 59 The next attested Roman military action was the so-
called Gaetulian 'war', which began perhaps in AD 3 and was
brought to a close by Cossus Cornelius Lentulus in 6. Though
the M usulamii seem also to have been involved, the campaign was
directed mainly against the Gaetuli, who, according to Cassius Dio
(55· 28. 3-4), refused to submit to the Romans60 Perhaps, as
Marcel Benabou has suggested, the Romans had tried to force
the Gaetuli to stay in certain areas or had intentionally blocked
their transhumant routes 6 ' It is diffieult to judge the seriousness
of the fighting, which appears to have extended along the whole of
the southern limit of Roman-controlled territory from Mauretania
to Leptis Magna: Florus describes it (2.31) as 'an uprising rather
than a war' (tumultuatum magis quam bellatum).
It was also under Augustus, probably in AD 14, that the Third
Augustan legion built a road to connect its winter quarters at
Ammaedara to the coastal town of Tacapae 6 ' It evidently cut
across the traditional territory of the Musulamii, which appears
to have extended from Sieca Veneria south to Theueste, and prob-
ably also across their transhumant routes. The road is sometimes
said, therefore, to have been the cause ofthe Musulamian rebellion
which began in AD 17, and which lasted, under the resolute leader-
ship of Tacfarinas, until 24. 63 There is, however, no evidence to
58 In~criptiones Italiae 3. 569; see also Fentress. Numidia, 65.
59 Plmy, HN 5. 36; see also Mattingly, Tripolitania 43 70.
60 The rebellion is mentioned also in Velleius Pater;ulu~ 2. II6. 2, and in Orosius
4· I 8, ~4; see also Le Bohec, La Troisieme Legion Auguste, 34.
6r Benabou: La Resistance africaine ci la romanisation (Paris, 1976), 65; but cf.
Fentress, Numidia, 65.
62 ~~nstruct.ion of the ro~~ is att~sted by CIL 8. 22786,22789; A. Merlin (ed.),
InscnptlOns larmes de la TWUfne (Pans, 1944), nos. 73-4; cf Whittaker, Frontiers, 44,
28 3 n. 16., The size ~nd location of the base at Ammaedara are unknown: V. A.
Maxfield, T~e Frontlers <?fthe Roma~ ~mpire: Some Recent Work', JRA 2 (19 89),
343· On the s1te, see espectally P. A. Fevner, Approches du Maghreb romain (Aix-en-
Provence, 1989-90), i. 106--8.
63 .~. Brett and E. W B. Fentress, The Berbers (Oxford, 1996), 46 ; Fentress,
NUmldlG, 66; I-M. Lassere, 'Un conftit "routier": Observations sur les causes de
The Roman Frontier-Zone 39
show that any land was taken from the Musulamii or that their use
of it was in any way restricted 64 None of the boundary-stones
(cippi) that records centuriation in their territorydates to the per-
iod bejare the rebellion. 6s Tacitus probably had !l generally nght
(Ann. 2. 52) when he located the origins of the insurgeney in the
raiding activities of the small band of former auxiliaries who fol-
lowed Tacfarinas (it is not, however, necessary to believe also, with
Taeitus, that they were deserters)66
The course of the war, which appears to have been joined by the
Gaetuli and Garamantes, and which seems at some point to have
spread from the Tebessa mountains around A~~aedara west
towards the Auros, has been desenbed many bmes: there 18 little
to be gained by re-examining it here. What really matters, I think,
is that the rebellion was never duplicated, either in scale or 1ll
length, in any part of Roman-era AI germ.
. 68
.
At least, so it seems. It has often been remarked that the anelent
sources have suspieiously little to say about armed resistanee to
Roman expansion in north Afriea. Little of what may have hap-
pened can now be reconstrueted 69 It is hard to believe, however,
la guerre de Tacfarinas', Ant. afr. 18 (1992), II-2S; M. Rachet, Rome el l~s Ber-
beres: Un probleme militaire d'Auguste a Dioclttien (Brussels, 1970), 75; Whittaker,
Frontiers, 79.
64 Cf. Daniels, 'The Frontiers: Africa', i. 238; MacKendrick, Nor~h African
Stones 216 who maintains that the rebellion was a product of MusulamJan resent-
ment ~t RO~lan 'urbanization and land-grabbing'. .
65 See also Fentress, Numidia, 67. On centuriation, see especial!y R. Chevalher,
'La Centuriation et les problemes de la coloni~ation rom~i~e', l!tudes, rurales, 3
(1961), 54-80; 1. Soyer, 'Les Centuriations romames en Algene onentale , An!. afr.
TO (r976), 107-80.
66 See also B. D. Shaw, 'Pear and Loathing: The Nomad Menace and North
Africa', in C. M. WeHs (ed.), L'Ajrique romaine: Les Conferences Vanier 1980.
Roman Africa: The Vanier Lectures /980 (Ottawa, 1982), 36-8. . ..
67 e.g. by Benabou, La Resistance africaine, 75-84; A. Berthle.r, La NUffUdze:
Rome el la Maghreb (Paris, 1981), 100-7; see also R. Syme, 'Tacfannas, the Musu-
lamii and Thubursicu', in P. Coleman~Norton (ed.), Studies in Roman Social a~d
Economic History in Honor 0/ A. C. Johnson (Princetoll, 1951), II3-3~. The mal11
Roman sources are Tacitus, Ann. 2. 52; 3. 20-1, 32, 73-4; 4. 23-6; Vellems Patercu-
lus 2. 125. 5, 129. 4; Aurelius Vietor, Caes. 2. 8.
68 So P. Gamsey, 'Rome's African Empire under the Principate', in P. Garnsey and
C. R. Whittaker (eds.), Imperialism in the Ancient World (Can~.brid~e, 1~78), 252.
69 Cf. Ravell, Rome in Africa, 59. A catalogue of north Afncan wars, ~o BC-A~
298, may be [ound in Benabou, La Resistan.ce.afr~caine, 250-1. Indeed, Benabous
review of the evidence is so thorough that 11 1S dlfficult to unders.tand wh~ D .. F.
Graf, JRS 82 (1992), 278, maintains that there was 'no compreh~ns.1ve exammat10n
of all the evidence from North Africa' before A. Gutsfeld, Romlsche Herrschaft
 4° The Roman Frontier-Zone
that arebellion as large and dangerous as that of the M usulamii
could have gone unreported. The long-held notion that indigenous
resistance to Roman occupation was ferocious and unremitting~
which goes back at least to Rene Cagnat's L'Armee romaine
d'Ajrique el l'occupation militaire de I'Ajrique sous fes empereurs
(19 13), must now be abandoned. There is simply no evidence to
support the view that native military resistance was tenacious.70
Perhaps the best indication of north Africa's widespread
acquiescence in Roman rule is the relatively sm all size of the
Roman military presence there (except in Mauretania)-a single
legion (the Third Augustan) and probably about 10,000- 1 5,000
auxiliaries, perhaps 20,000 men in all in the provinces of Africa
Proconsularis and Numidia. 7 ' And some at least of what moderns
have classified as native rebellions seem to be acts rather of
brigandage 7' (the line is, admittedly, an easy one to cross, and it
is impossible now to determine whether any of the so-called
'brigands' may have been motivated by nativist sentiment).
Serious and sustained military resistance seems to have been
centred father in Mauretania, especially [rom AD 40, when
Caligula ordered its annexation. The Moorish rebellion that his
po!icy inspired appears at some point to have involved also the
Musulamii, and to have spread even to southern Numidia. It
seems, tao, that armed resistance recurred sporadically for some
time after the rebellion was extinguished (AD 45), and perhaps well
into the second century.73
Outside Mauretania, there appears to have been little military
und einheimischer Widerstand in Nordafrika: Militärische Auseinandersetzungen
Roms mit den Nomaden (Stuttgart, 1989).
7° Pace Benabou, La Resistance alricaine, 248; cf. P. Leveau, 'Paysans maures et
villes romaines en Mauretanie cesarienne centrale (la resistance des populations
indigenes a la romanisation dans l'arriere-pays de Caesarea de Mauretanie)'
Me!anges d'arc!7eo!ogie el d'histoire de rEeole Fran('aise de Rome, 87 (1975):
857-7 1; M. SpeideJ, 'A.frica and Rome: Continuous Resistance?', Proceedings oI
the Alrican Clussical Association, J3 (1975), 36-8; and the exchange of views in
M. Benabou, 'Les Romains ont-ils conquis I'Afrique?', P. Leveau, 'La Situation
coloniale de l' Afrique romaine', and Y. Thebert, 'Romanisation et deromanisation en
Afrique: Histoire decolonisee ou histoire inversee?', Ann. ESC 33 (1978), 64-9 2.
7) See Daniels, 'The Frol1tiers: Africa', i. 235-6; cf Raven, Rome in Africa, 58.
2
7 MacKendrick, North Alrican Stones, 328.
73 Cf. Garnsey, 'Rome's Africal1 Empire', 252. The rebellion is described in
Bellabou, La Rhistance alricaine, 89-95; D. Fishwick, 'The Annexation of Maur-
etania', Historia, 20 (1971), 467-87; the main Roman sources are Suetonius Galba
7-8; AureJius Victor, Caes. 4. 2-3; Cassius Dio 60. 9. 6. ' 61-103- For Zarai (mod. Zraia), see Fentress, Numidia, 94.
The Roman Frontier-Zone 41
activity of any kind in the half century that followed the defeat of
Tacfarinas. We may suspect that there were localized hostilities
which have escaped notice. It is likely, too (but equally unproven),
that Roman military surveyors were busily engaged in cadastrating
the land that had been seized by the army. There is no indication,
however, of any significant change in Roman strategy or in the
positioning of Roman forces until the time of Vespasian, when the
legion was moved, probably in AD 75, from Ammaedara to
Theueste, with the resul! that the frontier was shifted a little
west and south. 74
The fron tier was moved again, much further to the west, in the
late 70S or early 80S AD, when a number of forts were COll-
structed in the northern reaches of the Aur"s mountains-at
Mascula, Aquae Flavianae, Vazaivi, and Lambaesis (where the
first of several Roman camps was buil! in 81).75 It may be
conjectured that forts were built also in the region of the Chott
Djetid. 76 Sometime later a road was built to connect Lamasba to
Zarai and, following a !ine north of the Hodna mountains, to the
military outpost at Auzia (mod. Souk EI Ghoziane, south-east of
Algiers)77 By the end of the century, the line of Roman fortifica-
tions stretched 400 miles from Tacapae to Mauretania (see Fig.
2·3)·
74 AD 75: Daniels, 'The Frontiers: Africa', i. 240. There is .no reason to. bclieve,
with MacKendrick, North African Stones, 218, that the legIOn was statlOned at
Mascula after it was moved from Ammaedara and before it arrived at Theueste.
It can hardly be agreed either, with Andre Berthier, La Numidie, 122, that Vespa-
sian's purpose in relocating the legion was to 'neutralize' the Aures mou~1tains,
which are more than 40 miles from Theueste. The legion seems to have remamed at
Theueste until AD IIS/I7, when it was trallsferred to Lambaesis: Le Bohec, La
Troisibne Legion Auguste, 362; cf. Raven, Rome in Alrica, 68.
75 For the forts, see Daniels, 'The Frolltiers: Africa', i. 240. Pace Y. Le Bohec, Les
Unites auxiliaires de l' armee romaine en Afrique Proconsulaire et Numidie sous le
Haut-Empire (Paris, 1989), 162, they did not 'encirc1e' the mountains. On ~ascula,
see Fentress, Numidia, 96; there has been IlO significallt excavation at the slte. For
Vazaivi, see T. R. S. Broughton, The Romanization of Afriea Proconsularis (Balti-
more, 1929), 103. Lambaesis: Fevrier, Approches du Maghreb romain, i. II 1. A pla.n
of the camp may be found in Le Bohec, La Troisieme ~egion Augu.ste, 363. The1:e lS
an excellent aerial photograph of the whole of the leglOnary base 10 MacKendnck,
North African Stones, 222-3. It used to be thought that the extension ofthe frontier
to the region west of Theueste belonged to the 2nd cent. An: see e.g. Broughton,
6
Romanization oI Africa, 104. 7 Mattingly, Tripol:'.tania, 77: 80.
77 MacKendrick, North Alrican Stanes, 219. On Lamasba (mod. Am Merwana),
see B. D. Shaw, 'Lamasba: An Ancient Irrigation Community', Ant. afr. 18 (1982),
 The Roman Frontier-Zone 43
The Frontier under Trajan and Hadrian
In keeping with the aggressive posture that he exhibited on Rome's
other frontiers, Trajan set out to expand the territorial limits of
Roman power a1so in north Africa. In Mauretania, the fron tier
was pushed west, with the road from Zarai to Auzia extended to
the edge of the Titteri range, and from there via Sufasar to the
Chelif valley7 8 But the principal military objective of Trajan's
African policy, it seems, was to encircle the massif of the Aures
mountains 79 A military colony was plan ted at Thamugadi (mod.
Timgad) in AD 100, and connected by road to the Flavian outposts
that had been established along the northern flank of the Aures80
Then, from about AD [04, the frontier-line was shifted south to run
along the southern edge of the Aures and Nememshas81 At the
direction of the legionary commander (legatus), L. Minicius N atalis,
a fort was constructed at Ad MaiDres (mod. Hr. Besseriani) in AD
[04 or 105, and linked by road to the co ast at Tacapae 8 ' The road
~ was later extended along the southern and western reaches of the
:az ~

i:' Aures to three military outposts that were established probably at

-0
ru
B
e "2;:: i'l
~
about the same time-Vescera, Calceus Herculis (mod. EI
ru ~
Kantara), and further to the west and north, in the broad plain
"
0 Cl
~
u
0
u
of the Hodna basin, Thubunae 83
"
~

~
0
Ce
~ It was probably also under Trajan, perhaps in AD [[5/17, that
] u

., "•" • 0
•E '"
~

""
0
~
• g.
':;
Il
~
0
-0 u

E <ii "~
ru
78
79
See Daniels, 'The Frontiers: Africa', i. 242.
So Broughton, Romanization oI Ajdca, 119; Le Bohec, La Troisieme Legion
~
u
:;;;" ~ "'
u
2 Auroste, 370.
'"
0
r- « ~ o Thamugadi is located in fertile, rolling country north' of one ofthe main passes of
] 'il
,. .... - ... ,.. ... ,.-"
~

~-)
I
'" r-:..oci 0, 0

ru
:ij
~

.~
~
the Aures. The site is described in Broughton, Romanization of Aji'ica, I J9; Manton,
Roman North Aji'ica, 98-9. There is 00 evidence of a pre-existing (indigenous)
settlement: Fentress, Numidia, 127-8. 81 Cf. Whittaker, Frontiers, 79,
,I •
0
2
·R ~
82 L. Minicius Natalis: eIL 8. 2478 (= 17969), 2479 (= 1797I). On the fort, which

.,
I
.co"
"
"
.~

iL
8
"-
«"-
""
~
u measured qo X 100 m., see Fentress, Numidia, 97; MacKendrick, North African
Stones, 243. For the date, Fevrier, Approches du Maghr{'b romain, i. I I I. The road is
I
I .~
"
~ .co •" described in Raven, Rome in Africa, 91; see also Daniels, 'The Frontiers: Africa', i .
I N '2 2 S
0
0
M
ci 242; Whittaker, Frontiers, 79.
... "\, .;l N" 0'
~
""
0 0
\",,"-
«
\ - ci cA "' "'.;- .;, ~
:::;
ob
i.i:
83 On Ca1ceus Herculis, whose ruins lie on the east bank of the Oued EI Kantara,
see Fentress, Numidia, 91 (except for a bridge that crosses the oued, no military
§
8
:E
-.
\
I

"(
\
\
constructiori has been found). The earliest epigraphic iodication of a Roman mili-
tary presence dates to AD q6-7: Le Bohec, La Troisibne Legion Auguste, 425. For
Thubuoae, see Fentress, Numidia, 92. On communicatioo-lines through the Aures,
see P. Morizot, 'Le Reseau de communications de la IIIe legion de Lambese au
\
Sah ara a travers l' Aures', in C. Lepelley (ed.), Actes du IVe colloque international sur
l'histoire ef l'archeologie de l'Afrique du Nord (Strasbourg, 5-9 avril, 1988) (Paris,
199 I),40 9-26 .
 44 The Roman Frontier-Zone
the legion was moved west from Theueste to Lambaesis where it
remained until 2 38 (when it was temporarily disbanded', perhaps
8 north, and Lambiridi (mod. Kherbet ouled Arif) to the south. 89
for twenty years). 4 Probably around the same time a veteran
colony was established at Diana Veteranorum, about 20 miles
east of Zaral and north-west of Lambaesis.85
The expanded military frontier-line established in the time of
TraJan seems to have been moved hardly at all under Hadrian.
Instead, forts were built (along the line north of the Auros for
example) to fiU gaps in the somewhat scattered Trajanic out;ost-
network, wlth the result that it was transformed into what Charles
Damels has called 'a regular frontier system,86 In Mauretania, the
more or less open frontier-road that ran from Auzia to Chelif was
protected by forts built west of Auzia at Rapidum and at Thanar-
am,usa Castra ,in AD 122, perhaps in response to continuing native
reslstance, WhlCh seems to have led to actual fighting in 118 and
agam m 122, and which appears to have been centred in the
Kabylie mountains north of Auzia and in the forbidding Ouarsenis
masslf south and west of Sufasar (what Benabou has called 'llots
de resistance').8 7 A few miles south and west of Vescera, and just
south of the Oued Djedi, a large mud-brick fort (it occupies 7. 25
acres) was constructed at Gemellae (mod. EI Kasbat), probably in
AD 125/6.88 The Batna mountains, which lie west and north of
84 Far the dates, see Le Bohec, La Troisieme Legion Auguste, 362. Cr. R. P. DUl1can-
Jones, Th~ Economyoft~e Roman Empire: Qu~ntitafiveStudies, 2nd edn. (Cambridge,
19 82), 67: th; cons,trUC!10n ofthe first camp m AD 81 'may not imply transfer ofthe
whole legIOn . J?amel~, The Frontiers: Africa', i. 240, takes a similar position. Cf. also
Raver:, l!0me In Alnca, 68-9: the legion was moved by Trajan 'ar his successor
H~~nan . M~r:ton, Ron:~n N~rth ~jr~'ca, 86, is wrong in dating the transfer to AD 128:
The ealhest survlVlng l11SCnptlOn from Diana Veteranorum (eIL 8. 45 8 )
dat~s to AD I4I; the Trajanic date is suggested by Broughton, Romanization ~f
Afl'lca, 134-5 n. 75.
:~ '~he Fron~iers: Afric~', i. 242; see also Raven, Rome in Africa, 9I.
. Be.nabou: ~es Rom~ms', 85. See also DanieIs, 'The Frontiers: Africa', i. 24 6.
It IS ~ntlr.eIy posslble that It was the construction ofthe camps which touched off the
fi~ht.mg 111 AD 122,. on which see Benabou, La Resistance africaine, 121-34. For
R<lpld~m,.see ~spe~tally J-P. Laporte, Rapidum: Le Camp de la cohorte des Sardes en
A!auretal1le Cesartel.me (Sassari, 1989); the second cohort Sardorum occupied the
slte from AD 122 ~ntJ12,08, when it was moved to Altava; the town was walled in 16 7;
see also ~. SpeI deI, The Roman Army in North Africa', JRA 5 (1992) 4°1'
MacKendnck,
88 ,
North
, v:. ,.
Al'rican Stones 250 ' ,
... ~or .the d,at~, s,ee Fe~nel~ Approc~les du. M~ghreb r?main, i. III; Daniels, 'The
FlOntlers .. Afr~~a: 1. 244, cr. Blr~ey,. Hadna111c frontler Policy', 29 (the fort 'is
shown by l?SClIptlOns to be Hadna111c'); Leschi, Etudes d'epigraphie 322 (AD 128)'
MacKendnck, North African Stones, 243 (AD 12617). On the fort: see Fentress:
The Roman Frontier-Zone 45
Lambaesis, were surrounded by several forts and settlements,
Lamasba and Lamsorta on their western flank, Lamiggiga to the
But the development that is most conspicuously Hadrianic-in so
far as it seems to symbolize a policy of retrenchment-is the
construction of the Jossatum.
The fossatum (a word little used in antiquity) is easily the most
puzzling of the Roman frontier fortifications in the region. Jean
Baradez, who in 1949 published what is still the most comprehen-
sive survey of its several parts (Fossa/um Africae: Recherehes
aeriennes sur l' organisation des confins sahariens a l' epoque
romaine), was convinced that its purpose (which is discussed
below) was to protect agriculturallands against the depredations
of nomadic raiders from the desert, to create, in his words, a 'zone-
tampon'.9 0 It was remarked earlier that his views, which were
adopted by many of his contemporaries, are today widely rejected,
in large measure because it is now dear that the Jossatum never
formed a continuous barrier or really much of a barrier at alL"
Baradez dated the system-the ditch itself and its associated
walls, towers, and gateways-to the time of Hadrian, on several
grounds: that the watch-towers which were part of the fossatum
just sonth of Gemellae are similar in design to those at the fort,
which is securely dated to the mid-120S; that they also resemble the
turrets at Hadrian's wall; tbat the ditch south of Gemellae is very
much like the one at Hadrian's wall; that the principal finds,
including coins and pottery, are consistent with a date in the
120S; and that linear barriers are characteristic of Hadrian's way
of thinking about fron tier defence9 ' His position has occasionally
Numidia, 83 (with bibliography). Pace Benabou, La Resistance afi'icaine, 73, there is
no reason to suppose that the Romans aimed at 'occupying' the Aures.
89 Broughton, Romanization oI Africa, 136. On 'Lam-' in Algerian pIace~names,
see Shaw, 'Lamasba', 64-5 n. 2, who suggests that it may be derived from a
paIaeoberber term for 'spring'.
9° Fossatum Afi'icae, 358. On Baradez, see especially D. Williams, The Reach 01
Rome: A History 01 the Roman Imperial Frontier Ist-5th Centuries AD (New York,
1996), 136-46.
9' See especially Maxfield, 'Frontiers', 343; Isaac, Limits 01 Empire, 414,
9" 'Complements inedits au Fossatum Africae', in Studien zu den Mililärgrenzen
Roms: Vorträge des 6 internationalen Limeskongresses in Süd-Deutschland
(Cologne, 1966),200-10.
 The Roman Frontier-Zone 47
been challenged, but it seems now to be generally agreed tbat be
had it right: tbefossatum is almost certainly 'Hadrianic'.93
It is divided into three parts (what Baradez identified as a fourth,
\
,, easternmost sector extending east from Ad Maiores towards the
I Chott Djerid has been shown by Pol Trousset to be a road):'4 one
"" ...... --, section of it, the Seguia Bent el Krass, is located just south of the
\
Oued Djedi and the Hadrianic fort at Gemellae; a second section
runs 28 or so miles from Mesarfelta to Thubunae, cutting across
south-western access-routes to the Aures mountains; a third, which
is both the longest (approximately 87 miles) and least studied of the
three, encircles the eastern end ofthe Hodna mountains (see Fig. 2-4).
The Seguia Bent el Krass seclion has been described briefly
already: about 37 miles in length, with its eastern end intersecting
the Oued Djedi, it consisted of a mud-brick wall fronted by a V-
shaped ditch 95 Baradez concluded that the wall was also dissected
by narrow gateways at mile intervals, and that it was provided with
towers placed singly about half-way between each set of gates. 96 It
has been suggested recently that the arrangement of the gateways
and towers may have been a good deal less regular than Baradez
supposed: there are, for example, far more towers than his scheme
would allow; some of them are grouped; and while some are on or
very close to the wall, others are located some distance away from
it. 97 But it may be that in its original (Hadrianic) form the Seguia
Bent el Krass was much as Baradez described it, and that its
irregularities are a product of post-Hadrianic modifications. 98
The configuration of the other two sections of the fossatum seems
to be much the same: a ditch of varying dimensions; a narrow wall
that in places gives way to a mound; gateways and watch-towers. 99

93 D. 1. Mattingly and G. D. B. Jones, 'A New Clausura in Western Tripolitania:

Wadi Skiffa South', üb. Stud. I7 (1986), 94. So, too, Birley, 'Hadrianic Prontier
Poliey', 29; Daniels, 'The Frontiers: Africa', i. 244; Le Bohec, La Troisieme Legion
Auguste, 371; Raven, Rome in Africa, 76-7; Whittaker, Frontiers, 48, correcting an
earlier view ('Trade and Frontiers' , I I 2) that the Jossatum was 'Trajanie or Hadria-
nie'. See also the diseussion in Fentress, Numidia, 98-101.
94 Baradez: Fossatum AJricae, 109-11, 146. Trousset: 'Signification d'une
frontiere'. 95 See also WjJJiams, Reach, 135-6, 144-5.
96 Fossatum Africae, 93-108. The Seguia Bent el Krass sector is described also in
Lesehi, Etudes d'epigraphie, 37-8; Le Bohec, La Troisieme Legion Auguste, 370;
there is a good photo in Leschi (at 36). The gateways are diseussed also in Fentress,
Numidia, 114. 97 Daniels, 'The Frontiers: Afriea', i. 244.
8
9 See Maxfield, 'Frontiers', 343.
99 See also Daniels, 'The Frontiers: Afriea', i. 244-6.
 The Roman Frontier-Zone
In their component parts, the sections of the fossatum are
broadly similar to the somewhat better preserved, and probably
somewhat later, linear barriers that have been recovered in Roman
Tripolitania (mainly in southern Tunisia)-the so-ealled clausurae,
most of whieh (twelve have now been identified) eonsisted of a
diteh, wall, wateh-towers, and gates, wo The Gebel Tebaga c/ausura
south-west of Capsa (mod, Gafsa), for example, whieh was built
across about 10 miles ofplain-Iand between the Gebe! Tebaga and
the Gebel Ma.tmata, consisted of a stone bank fronted by a ditch
and accompanied by watch-towers, with a single gate near its
south-east end. '0' The best preserved of the c/ausurae is at Bir
Oum Ali, a narrow defile in the Cherb mountains about 35 miles
south-east of Capsa. An imposing mortared wall with two-faced
masonry, about 4.5 feet wide and almost 20 feet high, dated, it
seems, by its associated pottery sherds to the late second or early
third century AD, it looks remarkably like Hadrian's wall. M
The Developed Frontier (ta AD 235)
The Hadrianic frontier-system was changed very little in the sixty
or so years that separate Hadrian's death from the accession of
Septimius Severus in AD 197. Under Antoninus Pius, in AD 145, the
road that linked Thamugadi to Lambaesis seems to have been
extended to Vescera, completing the encirclement of the Auros
mountains begun under Trajan. 1 0 3 Two mid-century inscriptions
attest other military operations in the region. One (CIL 8. 24 65 =
17953), from Mena'a, in a valley ofthe central Auros that was horne
to a much-used transhumant route, records that a detachment of
100 See especially Mattingly and Jones, 'A New Clausura'; see also O. Brogan,
'Radd Rajar, a "Clausura" in the Tripolitanian Gebel Garian South of Asabaa'
Lib. Stud. Ir (1980),45-52, on the easternmost of thc clausurae. Mattingly, Tripo~
Utania, 79, 115, argues that at least some of them are pre-Severan. They are mappcd
in Whittaker, Frontiers, 80. 101 Mattingly and Jones, 'A New Clausura', 89.
102 See also Mattingly, Tripolitania, 108; Mattingly and Jones, 'A New Clausura',
88-9, 95· Photograph in MacKendrick, North African Stones, 60. It appears that
most of the other clausurae were originally not much more than 6---9 feet high:
Mattingly, Tripolitania, 113.
10
3 Construction of the road from Lambaesis to Vescera is thought to be
attested by CIL 8. 10230, which records the presence of a detachment of the
Sixth legion Ferrata: Fentress, Numidia, 114; see also Raven, Rome in Airica, 68.
Rowever, the inscription may refer instead to local surveillance operations; see
below, p. 56.
The Roman Frontier-Zone 49
the Third Augustan legion was morans in procinctu C'waiting in
readiness').'04 The second (CIL 8. 4322), from Casae, just north
of Lambaesis, mentions a vexillation of legionaries morantes ad
fenurn sec(andum) ('waiting to mow the hay'). It has been suggested
that the soldiers may have been protecting local agricultural
labourers (but against what?), or, and what is more likely in my
view, that they themselves were harvesting hay or overseeing its
collection. lOs Later in the second century, under Commodus, a
military observation-post (burgus speculatorius) was established
on the route between Calceus Hereulis and Mesarfelta. w6 To the
west, in Mauretania, a fort was constructed in AD 148/9 near Med-
jedel, on the northern slopes ofthe Ouled Nail mountains, about 70
miles due south of Auzia, and almost 180 miles west of
Lambaesis. W7 Its purpose is not readily apparent; perhaps it was
intended to serve as a forward survey post for the frontier-road that
lay far to the north beyond the High Plains and the Titteri
mountains. 108 Later, under Commodus, the road itself was
provided with wateh-towers (between Auzia and Rapidum). W9
It appears, then, that much of what was aecomplished under the
Antonine emperors was in the nature of housekeeping (the excep-
tion is the fort built near Medjedel). The Hadrianic fron tier was
not restructured in any significant way until after the accession of
Septimius Severus, when the !ine of Roman fortifications was
pressed to the very edge of the deser!. 'w
A ehain offorts, including EI-Gara and Aln Rich, was built west
of Gemellae at least as far as Castellum Dimmidi, where a fort,
10
4On Mena'a, see P. Morizot, 'Vues nouvelles sur I'Aures antique', CRAI
10
(197~), 309. 5 See also Fentress, Nwnidia, 109.
10 CIL 8. 2495; Daniels, 'The Frontiers: Africa', i. 250.
10
7 See Daniels, 'The Frontiers: Africa', i. 248.
lOS Three previously unpublished forts which are described briefl-y in P. Salama,
'Quelques incursions dans la zone occidentale du limes de Numidie', Ant. afr. 27
(1991),93-105, and which are located in the region east ofMedjedel, appear also to
date to the time of Antoninus Pius.
10
9 Daniels, 'The Frontiers: Africa', i. 250.
110 At about the same time, the frontier in Tripolitania was also advanced to the
desert with forts established at Ghadames, Gheria el Garbia, and Bu Njem: see P.
Trous~et, 'De la montagne au desert Limes et maitrise de l'eau', ROMM 4.1,-2
(I986), 93; on the Tripolitanian frontier in general, see Rebuffat, 'La Frontlere
romaine', 225-47; R. G. Goodchild, 'Oasis Forts of Legio III Augusta on the
Routes to the Fezzan', Papers ofthe British School ai Rome, 22 (1954), 56-65; the
frontier is mapped in Fevrier, Approches du Maghreb romain, i. I63. See also Isaac,
Limits of Empire, 126, on Severan expansion in Arabia.
 The Roman Frontier-Zone 51
half a hectare in size and designed to house 500 cavalrymen, was
constructed in AD 198 at the direction of the legionary commander,
Q. Anicius Faustus.''' lt has been suggested that the frontier-line
may have been extended even further to the south and west to
\
,, Laghouat, south-east of the Djebel Amour (see Fig. 2.5). m But
I nothing identifiably Roman has been recovered there.
"'" ... --, \ There is no indication either that the Romans systematically
explored the desert beyond Castellum Dimmidi. Only two Roman
armies are known to have penetrated the western Sahara. In AD 4 1
or 42, Gaius Suetonius Paulinus, pro-praetorian commander in
Mauretania, led his troops into the desert in pursuit of some
Moorish insurgents; the Romans turned back after ten days'
march through 'deserts covered with black dust' (solitudines nigri
pulveris) and 'places uninhabitable by their heat' (loca inhabitabilia
jervore).' '3 Around the same time, his colleague, Gnaeus Hosidius
Geta, pursued the Moorish chieftain Sabalus into the deser!. "4
There are no reports of subsequent Roman expeditions into the
region. The record is probably incomplete. But it can safely be said
that the Romans had very little interest in the deser!.
A second Severan network of forts was constructed along a line
that runs north ofthe Ouled Nail mountains, at el Guelaa and Bou
Saada in the region east of the Antonine fort near Medjedel, and,
] west of it, at Korirein, Aiu el Hammam, and Zenina. Roman
c
0
0
remains have been found even further to the south-west, at Agneb,
c el Bayadh, and el Khadra."5
E ~ lt is difficult to discern a military purpose in all this construc-
'§
\. 0 ~ tion, the effect of which was to encircle the Ouled Nail mountains.
-,,
r~"- __ ,.. .....
a
&
~
~

"i
~
~
111 On EI-Gara, see Fentress, Numidia, 87 (with bibliography). The date of the
0 eS
construction of Castellum Dimmidi is attested by AE (I948), 214; see also Leschi,
] ·S
~
Etudes d'epigraphie, 322; cf. Raven, Rome in Africa, 78. On the fort, see especially G.-
0 .~ e. Picard, Castellum Dimmidi (Paris, 1957). Jt was occupied until AD 238: Mac-
"
.5
~
0
ci:;
Kendrick, North African Stones, 246.
113 Pliny, HN 5. 14-15.
112 Fentress, Numidia, 114, 122 n. 23·

8" ~ J 14 Cassius Dio 60. 9. 1-5. On Roman exploration of the central Sahara, which
0.
.!( &:: was in part a by-product of punitive measures directed against the Garamantes, see
.;.. Raven, Rome in Africa, 62-3; on trans-Saharan trade in the Roman era, see
~ oi especially 1. T. Swanson, 'The Myth of Trans-Saharan Trade during the Roman

J """"
~ ~
Era', International Journal of African Historical Studies, 8 (I975), 583-600; M.
Milburn, 'Romans and Garamantes: An Enquiry inta Contacts', in D. 1. Buck
and D. 1. Mattingly (eds.), Town and Country in Roman Tripolitania: Papers in
Honour of Olwen Hacket! (Oxfard, 1985), 241-61.
115 See Daniels, 'The Frontiers: Africa', i. 253·
 52 The Roman Frontier-Zone
Perhaps we are to understand that the mountains were a source of
indigenous resistance (there is no evidence for it). "6 The arrange-
ments cannot be said to have had a strategie purpose, at least not one
that can be interpreted as having been directed against an extern al
threat (from the south-west?), for their effect was greatly to
lengthen both the frontier-line and the salient that was formed
of the High Plains where they were squeezed between the forts that
ran along the northern slopes of the Ouled Nail and, to the north,
the line ofSeveran forts (described below) that followed the southern
edge of the Tell Atlas. It might be conjectured that the Romans
intended to encircle the High Plains by driving a line of fortifica-
tions across them, from el Khadra, say, to Cohors Breucorum, and
that their plans were, for one reason or another, never realized (it is
less likely, I think, that there are undiscovered forts in the western
reaches of the High Plains).
A third network of Severan forts (including Aln Grimidi) was
constructed along a line that ran parallel to the northern edge of
the High Plains, south of the Biban and Titteri mountains in the
east, and of the Ouarsenis massif and Frenda mountains in the
west. At about the same time, the frontier-road west of Auzia was
extended all the way to the region of the Oued Tafna, where forts
were established at Altava (in AD 202) and at Numerus Syrorum,
just 40 miles from the present-day border with Morocco. "7
The effect of the Severan arrangements in Mauretania was to
create a 'double fron tier', in which the Trajanic frontier-road west
of Auzia, which coincided with the 400 millimetre isohyet-the
theoretical southern limit of profitable cereal culture without
irrigation-and which continued to be thought of as the main
fortified line (praetentura), was now complemented by a second
array of fortifications well to the south in the steppe-land of the
High Plains (what milestones in the region call the 'new frontier-
line', nova praetentura). Ir8 Much the same can be said of the
frontier-system in east-central Algeria: an inner (northern) line
of fortifications, which connected Theueste in the east to Zarai
I!6 cr. P. Leveau, 'L'üpposition de la montagne et de la plaine dans l'historio-
gr~phie de,l' A~rique du ~ord anti9ue'" Annales de geographie, 86 (1977), 201-5.
7 Damels, The Frontters: Afnca', 1. 254.
118 P. Salama, 'Les Deplacements successifs du limes en Mauretanie Cesarienne'
in Fitz (ed.), Akten des XI internationalen Limeskongresses, 578; Trousset, 'De 1~
montagne an desert', 93; Whittaker, Frontiers, 81.
The Roman Frontier-Zone 53
and Auzia, tracked, in Susan Raven's words, 'pretty exact1y', ~he
400 millimetre isohyet; an outer (southern) system of forts, whlCh
extended from Castellum Dimmidi in the west through Gemellae
and Ad Maiores to Tripolitania, followed a line that ran between
the 100 and 150 millimetre isohyets, where the semi-arid lands that
can support the dry-farming cultivation of cereals glve way to the
desert. II9
The Military Presence
It seems now to be widely agreed that the Roman frontier in north
Africa was usually 'one of low priority'. 120 The e:idence ~s the
relatively small size of its occupying army. Only a smgle leglOn-
the Third Augustan-of perhaps 5,000 men was permanently
stationed in the Maghreb over the whole of the Roman era. On
the generous assumption of a complementary force, of about
15,000-20,000 auxiliary soldiers, Roman north Afnca s. approxl-
mately I 10,000 square miles can be said to have been garnsoned by
° .
an army f Just 20,000-25,000 men.
I2r
119 Raven: Rome in Africa, 78; cf. Trousset, 'De la montagne a~ ,~esert'; ?3;
MacKendrick, North Ajhcan Stones, ~4I. See also, Le Bohee, ~a T':01slCm~ LelflO,n
Auguste, 578, who distinguishes four dlserete ~rontJer-systems ~ s~ste!nes ?efenslf~ ~
in north Afriea not inc1uding the Mauretaman double-frontwr. a systeme aU,r~
sien' consisting' of the legionary base at Lambaesis, garrisons at Mena'a, VaZalVI,
Mas~ula, Casae, Zarai, Caleeus Hereulis, Gemellae, ,and the m?dern EI-Arou~ and
Hr. Sellaouine, two police-posts (burgi speculatorii) 111 the Aures, two ea,mps 111 ~he
. [ El-Outaya and the Mesarfelta-Thubunae segment of the jossatum, a
reglOno, " h'h' IddG He
'systeme saharien' in (what was to beeome) .NUt?l?ta, w IC me,u e ,eme a ~
Ad Maiores, EI-Gara, A'in Rieh, Castellum Dlmmldl, an? the SegU1a ~ent el Kra~s:
a 'systeme saharien' in western Tripolitania, e~mpnsed of garr,lsons at .,Sldl
Mohamed ben A'issa, Ksar Rhelane, Remada, and SI Aoun, and four hnear-?arIl,ers,
at Djebel Tebaga, Djebel Melab, Oued Skiffa, and Oued Zrala; .and a system,e
saharien' in eastern Tripolitania, which inc1uded forts at Zel~a, Bu NJem, Gasr Zerzl,
Gheria el Gharbia, Gheria el Sehergia, Aln Wif" Aln el-Aven.la, and Ghadan:es. I find
it hard to believe that Roman military strategy 111 north Afnca was as eoheslve as Le
Bohee's seheme makes it out to be. . . ", .
120 The quotation is from Daniels, 'The Frontlers: Afnea, 1. 235, see also Raven,
Rome in Africa, 58. ' , .
121 Daniels, 'The Frontiers: Afriea', i. 235-6, puts the .slze of the army at 3°,500,
his estimate of 24 500 auxiliaries is, in my view, tao hIgh, The figure for squar~
miles, a grass esti:nate of ~he pl~nar or 'fiat-map: area of the Rom~n-co~trolle.
territory in north Afriea, IS denved from ~haw, ~u~onomy and !nbute , ,68 ~lt
eombines his figures for Africa Proeollsulans, NumldIa, Mauret~ll1a Cae,s~n~nsl~,
and Mauretania Tingitana). The distribution of the north Afncan auxIlmnes IS
plotted in Le Bohee, Les Unites auxiliaires, 194-5·
 54 The Roman Frontier-Zone The Roman Frontier-Zone 55
Almost half of the soldiers were stationed m Mauretania could have been intended to achieve any of thhese purtPloseds, {~~
Tingitana (one of the most heavily armed of all the Roman , r h th wIll not ave ou Ive
provinees), '" Most of the rest appear to have been eoncentrated ~~~~~a~~Pt~~%se~~e~c~~7:~;who~~re likely been in their t~,~ave
in the region around Lambaesis and in the relatively narrow
fron tier-zone of Mauretania Caesariensis. In the absence of any mi~~~~s I~~~eO~~~?i~~e~ t~:~e~~~e t~;c~:~~e~~utine ofhthe ~ost;.
kind of data, it is impossible now even to estimate the size of the . . the two centunes from teen 0
indigenous population of the fron tier-zone, But it may safely be conquest frontIer army m h . ation of Severus
T f rinas' rebellion in AD 24 to t e assassm. .'
eonjeetured that the 10,000- I 5,000 Roman soldiers who were ac a d ' eriod in which there was liltle senous fightmg,
posted there were a signifieant proportion of the loeal population, Alexan er m 235, a p , '" f h 'us army
, Pf to think that the responslbIlItles 0 t e vano
rtappears, too, that their distribution was ehanged very Iittle in the lt I,St temay Ibnegl'ndicated by their location, But if the inscnptlOn that
period after the legion was moved from Theueste to Lambaesis (AD um s
records its list m ,
of tariffs had not been dlscovere d , noeon I
, suspeet t cl
II 5/17?), Many, perhaps a small majority, of the soldiers were sta- Id have imagined that the unit stationed at Zaral was expec e
wau . I28
tioned at ornear Lambaesis, Most oftherest, it seems, were posted to
to colleet customs dutJes, "d t' t from
the fortified frontier-Iine in Mauretania, to its advanee line of south- I ' I'k I too that the role of many umts vane , no JUS
ern outposts, or to one of the two series of forts that lay north and t IS
areaI etoy, another
, but also over tIme., "9 What IS reasonably
.
south ofthe Ouled Nail mountains, It is not unlikely either that small ~~:r from its wide~pread distribution across the frlonlIer-Zt~:~
legionary garrisons were established in the Aun,s mountains, per- and ~t several siles weil to the north of the fronlIer- me, IS , I
haps espeeially at what Michel Janon has called their 'points ou des
:~:i:;t a~d pn:~r~~f
m was never intended solely to defend the provmcla s
zones de passage', >23 The idea has been ehallenged by Elizabeth e':,ternal attack. lt functioned also, per;aps
Fentress, mainly on the grounds that only a single legionary post as an internal security force; it was, in Isaac s wor s, an ar
has ever been identified in the mountains-a garrison, probably t nd occupation'. 130 • .
smalI, at Ain el Aouad, whose presence is attested by a legionary conques ab ' t d on the basis of duty-rosters and slmIlar
It may e conJec ure , E that a
stamp on a water-pipe (fistula), "4 But it needs also to be remarked
that there has been little systematic exploration of the Aures,
It might be supposed that the veteran colonies scattered aeross
~~~~~d;e:~c~;~~:ds~:d~!~:,r t~:~;e~i~~~~::7~~~~;:t~~~E~:;:~~:
g
the frontier-zone (Thamugadi, Diana Veteranorum, Sitifis) were ities
. of camp-hfe; secunn 0 foo
Casae waItmg w' the h'ay'?)'
. , rebnilding structures;
t mo
meant to assist the legionary and auxiliary soldiers in providing anes near eqmpmen
repairing , t . 131 lt is likely also that many soldiers, and
for the region's seeurity or that they functioned as strategie reserves
ofmanpower which could be ealled up in emergencies, "5 So Robert
12
Broughton imagined that their purpose was to 'watch' the 7 A similar point is made in Isa.ac, Limits 0/ Empire, at 312-13.
natives, '26 It is hard to believe, however, that the veteran settlements See also Isaac, Limits 0/ Empire, 1.~4, 409 'die' in Maxtield and Dobson
128
12
9 M. Janon, 'Remarques sur la fronttere d e N uml ,
(eds.), Roman Frontie~ Stu~ies 19 9, 4 4i<. dd h L'Algerie dans I'Antiquite, 2nd
8 8
]22 See E. Fn':zouls, 'Rome et la Mauretanie Tingitane: Un constat d'echec?', Ant. 13° Isaae: Limits o} Emplre'd 2. Cf:bM, 't aas aane 'i~strument of domination'.
1'
'.11 R W Davies Service in the Roman Army, ~. .
aj;', 16 (1980), 65-93; Shaw, 'Autonomyand Tribute', esp. at 68. 82) 137 who esen es 1 d D
edn.1 (A glers, 19. , ,
U3 Janon: 'Recherehes a Lambese', 199. 13 On camp-hfe, see espeCJa y .. ' . L J F Keppie 'Armies on
Breeze and V. A. Maxfiel? C~e~ ~~~fie~~8;~iriob~0~ (~ds.), Ro'man Frontier
12

4 Fentress: 'Forever Berber?', Opus, 211 (1983), 167. For the water-pipe, AE
(I976), 716. Frontiers: Myth and ReahtJes, m British Museum Papyrus 2 51 ('Hunt's
8 22 8
/25 See e.g. Manton, Roman North Alrica, 86-7; W S. Hanson, 'Administration, Studies 19 9, 455-7· Casae: eIL 8. 43 . iers stationed at Stobi in Moesia in AD
Urbanisation and Acculturation in the Roman West', in D. Braund (ed.), The Pridianum') re.cords th~t som~ of the( s~~~ablY on military land). For the papyrus,
Administration althe Roman Empire (241 BC-AD 193) (Exeter, 1988),53-4. On Sitifis 100/5 were asslgned to de~~~ ero~s Preis on Papyrus (Cleveland, 197 1), no. 63;
(mod. Setif), see Broughton, Romanization 0/ A/rica, 126; Garnsey, 'Rome's African see R. O.
Elton, Fink, Roman
Frontiers, lltar~a so e~m,
115- I 6; see k 'Hunt's Pridianum: British Museum
Empire', 231. 126 Romanization 0/ Africa, II7. Papyrus 2851', JRS 48 (1958), 102-16.
 The Roman Frontier-Zone
perhaps espeeially the auxiliaries, were routinely employed in
patrolling and policing the countryside, and in the Roman
equivalent of ~immigration contro!'. 13 2 At a fork between two
roads near the southern end of the gorge of the Djebel Selloum
on the route between Calceus Herculis and MesarfeIta, a police-
post (burgus speculatorius) was established in the time of Com-
modus; commanded by a centurion and manned by a numerus
Hemesenorum, it is said to have provided 'new protection for the
safety of travellers' (ad salutem commeantium nova tutela). '33 It is
not unlikely that similar posts were set up across the whole of the
frontier-zone.
Soldiers of the Sixth legion Ferrata, whose presence in the
gorges of Tighamine in the Aures mountains is attested by an
mscnptlOn (CIL 8. 10230), are likeIy to have been engaged in
police or surveillance operations. '34 The recently published
ostraka found at Golas (mod. Bu Njem) in Tripolitania, which
consist mainly of daily duty-reports and personalletters written by
the members of an unidentified unit of numeri that was stationed at
several pre-desert outposts in the period AD 253-9, refer to sur-
veillance operations and to what might be called 'passport'
checks. '35 Fentress bas suggested that an inscription from Agneb
(CIL 8. 1567), south-west of Castellum Dimmidi, which attests the
presenee of legionary vexillations and two auxiliary cohorts in AD
I7 4, can be taken to indicate that the units were engaged in
patrollmg the northern reaches of the Saharan Atlas. '3 6 Regular
1}2 Davies.' Service, 56-8; on north Africa, see especially Rebuffat, 'Au-deli des
camps romalUs' , 490-2. On the functions of the auxiliaries, see also N. Benseddik
Les Troupes auxiliaires de ['armee romaine eil Mauretanie Cesariellne sous le Haut~
Empire (Algiers, 1982); Le Bohec, Les Unites auxiliaires
1}3 CIL 8. 2495; see also 2494 (::: ILS 2636); AE (1926), 145; (1933), 45 (revising
eIL 8. 2496~; (1933), 46. The site is described in Fentress, Numidia, 90-1; see also
114· <?f. Davles, Service, 60, who concludes that it was part of a network of similar
~urveIllance-posts whose purpose was to 'control movement and communications'
lU the area.
134 Cf. Fentress, 'Forever Berber?', 167, who believes instead that they were
probably there to build a road.
13~ The ,ostraka are published in Marichal, Les Ostraca; earlier notice in
Mancha!, ~es Ostraca. d~ Bu Njem', CRAI (1979), 436-52; discussion in J. N.
Adams, Lahn and PUllIC.lU Contact? The Case ofthe Bu Njem Ostraca', JRS 84
(I994), 87- 112 . On surveJilance operations, see Mariehai, 'Les Ostraca', 450; Les
Ostraca, 106-14.
136 Numidia, 114. The inscription may be thought to indicate also that there was
a fort at the sire: Daniels, 'The Frontiers: Africa', i. 253.
The Roman Frontier-Zone 57
soldiers might also be employed to guard private estates (against
bandits?). '37 They were sometimes used also to police market-
places: in the civilian settlement at Lambaesis, two standard-
bearers (signiferi) and their assistants supervised the operation of
the local market. '38
Judging from what seems to have been the experience of other
provineial armies, it might be expected that the soldiers stationed
on the Algerian fron tier were routinely engaged in eivilian engin-
eering and construction projects. '39 The orthogonal plan of the
civilian settlement at GemelIae has been said to indicate that the
army was responsible for its construetion or planning (or both). '40
But there is no real evidence of direct military involvement. Its
plan shows only that those who built it may have been infiuenced
by military styles of town-planning and construetion.
Mueh the same can be said of the similar eivilian settlement at
Ad Maiores. '4' And while legionary bases, like those at Theueste
and Lambaesis, invariably attracted indigenous merchants and
suppliers, at least some of whom eleeted to take up residenee
nearby, at first in the so-called tent-towns (canabae) which sprang
up around the camps, and later in the more permanent settlements
which grew out of thern, 142 there is no indication that the army
actively assisted in their planning or construction.
My impression then is that the Roman army in north Africa
built mainly for itself. '43 It will be remarked that the concIusion is
137 CIL 8. 14603 (= ILS 2305), the saltus Philomusianus, mid-Ist cent. AO; see
also Garnsey, 'Rome's African Empire', 347 n. 48.
8
13 eIL 8. 18219 (= ILS 2415); see also R. MacMullen, Soldier and Civilian in the
Later Roman Empire (Cambridge, Mass., 1963),59,92; B. D. Shaw, 'Rural Periodic
Markets in Roman North Africa as Mechanisms of Social Integration and Contra!',
Research in Economic Anthropology, 2 (1979), 103; 'Rural Markets in North Africa
and the Political Economy ofthe Roman Empire', Ant. afr. 17 (1981),56.
139 Davies, Service, esp. at 64. On the army's role in the planning and construction
of towns in post-conquest Britain, see B. C. Burnham, 'Pre-Roman and Romano-
British Urbanism? Problems and Possibilities', in B. C. Burnham and H. B. Johnson
(eds.), Invasion and Response: The Case 0/ Roman Britain (Oxford, 1979), 255-72,
esp. at 267; 'The Origins of Romano-British Small Towns', Oxford Journal of
Archaeology, 5 (1986),185-2°5; Hanson, 'Administration', 62; M. Todd, 'The Early
Cities', in Todd (ed.), Research on Roman Britain, 87; Wilson, JRS 82 (1992),292; for
Gaul, see E. M. Wightman, 'Military Arrangements, Native Settlements and Related
Developments in Early Roman Gaul', Helinium, 17 (1977),105-26.
'4° Fentress, Numidia, 134. '4' Cf. Le Bohec, Les Unites auxiliaires, 17I.
2
'4 See Broughton, Romanization 0/ Africa, 120; Raven, Rome in Afric~, IO~-I.
143 All of the 'military constructions' which Fentress, Numidia, 165, IdentIfies
were built either for military purposes or at military sites Iike Lambaesis.
 58 The Roman Frontier-Zone
constructed on mostly negative evidence But 't . . .
the I ' . I IS eonslstent wlth
h arger plcture that will be described over the next several
c apters, m WhICh the army of the fron tier-zone may fairl be
descnbed as Isolated from the local population. '44 y era: it cannot be disputed, for example, that such an arrangement
THE PURPOSE OF THE FRONTIER-SYSTEM145
That part of Roman-era Algeria which lay between the inner and
outer (that IS, northern and southern) frontier-lines is what lean
DespOIS and, after hirn, Whittaker have called 'the Waiting Zone'
~here, :11 Whlttaker's view, the Romans aimed at controIlin'
nomadlc a~d sem~-nomadic movements between early summe;
:hen there IS suffiCIent grazing in the zone to support flocks whil~
. aIvest labour filters northwards, and about JUDO to August whe
It becomes essentIal to get the flocks and herds northw;rds t~
Pastujre on the stubble'. The main funetion of the inner frontier-
I me, le goes on to say 't
"Wai . " , was 0 control the pastoralists in the
. . tmg Zone and regulate the very important matter of the
tJmmg of then entrance and exit from the zone-not too earl
(especlally m bad years) to trample the crops with their herds bU~
early enough to provide the harvest labour that was hired i~ th
. southern marche ~ ' . I46 A Ccord'mg to this reconstruction of the UT-e
pose of the frontIer, then, the principal objective of Roman p~iey
m north Afnea was to regulate the timing of the movements of
pastorahsts and their animals.147
'44 S
ee also B. D. Shaw, 'Soldiers and Society- The Arm i N ' . ,
(19 83),144. Cf. Elton, Frontiers 74' 'the high d ' f" y n. umldm, Opus, 21I
and the population of the area 'is o~e of th d ~r~e 0 ~ntegrah.o~ between the army
,145,Modern opinion is reviewed in e e llmg c ara:tensttcs o~.the frontier.'
d ~~nque~, CRAI, (199 0),5 65-80. M. Euzennat, La FrontIere romaine
Whlttaker: Land and Labour in N th Af·' . that thejossatum (or the clausurae) was essential to the regulating
Whittaker, 'Trade and Frontiers' 11' o,r nea, Kh~, 60 (1978), 349-50. Cf.
labourers aeross the borders befor~ th/ha:areful regulatIOn? to allow seasonal
to enter only after the erops had b th vest, but to permlt transhumant herds
system'. A similar purpose is attri~~~:; toe~~d,.p:ObfblY ~xplain the d~uble frontier
Tripalitania,37-8 79 225 221' D J M tf ~ npo ltaman. c!ausurae m Mattingly,
An Arehaeologic;l R~vie~' JJ?S 8" ( a )mg y and R" ~. Hltehner, 'Roman Africa:
147 Whittak F: .' 5 1995,174; cf. Manchal, 'Les üstraea' 448-51
er, rontzers, 91" see also A Berth" 'N·· ' "
Nomades ou sedentaires?' Buil,r" d' h,'1 " le:-, leIbes et Suburbures:
t
Daniels, 'The Frontiers: Afriea'~ 2 arc eo agle "algerienne, 3 ~I968), 293-300;
z
Rame in Africa, 78" Trousset 'Sign"fi 3;~ 44d~ 246 , Elt?,n, FrontIers, 103; Raven,
tiere', 62" ' , I ca lOn une frontIere', 935; 'L'Idee de fron-
The Roman Frontier-Zone 59
The theory is a seductive one, in part because it coincides neatly
with what is known of indigenous uses of the land in the Roman
would have benefited both the transhumant tribes, who could hire
out their labour at the grain and olive harvests, and the sedentar-
ists of the Tell, whose harvested fields would receive needed
manure [rom the pastoralists' animals" I48 But it is not, I think, a
compelling explanation of the Roman frontier-system, for it fails
to answer an important question: how was it in the army's or, for
that matter, the emperor's interest to regulate the seasonal
movements of the transhumant tribes? It seems to me that any
explanation of the purpose of the frontier-system must be deemed
unsatisfactory if it cannot be demonstrated that the execution of
that purpose will have served the interests of those who built and
maintained the fortifications, the army or the imperial govern-
ment. The idea that the frontier was designed to regulate the
interrelationship of transhumance and agriculture rests on an
implicit assumption, which is neither demonstrably eorrect nor
inherently likely to be true, that the emperors (or more broadly,
the Roman government) believed that they had a duty to protect
the interests of provincial, and in this case north African, agricul-
turalists, or, more widely, to regulate the economic and ecological
interdependence of sedentarists and pastoralists. '49
There is a second objection. All of our surviving evidence indi-
cates that the pre-Roman indigenous population of Algeria prae-
tised both agriculture and transhumance (sometimes, it seems, side
by side). '5° 1s it reasonable to suppose that they were unable to
regulate pastoralism before the coming of the Romans, that is,
before the Roman army intervened with its complex network of
roads, barriers, and fortifications? If we start with the assumption
of the interrelationship of agrieulture and transhumance, we must
inevitably find ourselves in the position of arguing that the native
population could not easily have managed without it. The 'double
frontier' may have had the practical effect of controlling seasonal
148 See Fentress, Numidia, 186; Raven, Rame in Africa, 77·
149 Cf. Isaac, Limits of Empire, 334, on imperial building.
IS° Fentress, Numidia, 19.
 60 The Roman Frontier-Zone
transhumance, hut it is hard to believe that it was eonstrueted for
that purpose,
For mueh the same reasons, it is unlikely that the fron tier-system
was deslgned to protect agncultural lands (in the words of one
study, 'proteger I' Afrique utile') either from the damage that might
be done to them by ammals or against raiders from the Sahara. '5'
But the notIOn seems not to be easily abandoned. A survey of
recent work on the Roman fron tiers concludes that the linear
barners of north Africa played 'a localised role in the control of
populatIOn movement (of seasonal transhumance, for example)
and In the protection 0/ areas 0/ cultivation'. 15 2
The idea has been given what may be its fullest expression by
Ene Blrley, aceording to whom the Hadrianic fron tier was
mtended to mark 'the limit up to whieh economic development
was to be permitted and indeed encouraged'. '53 Three objectives,
he suggests, mlght be met: first, and most importantly, the devel-
opment of farms along the frontier would produce 'food far the
frontter garnsons and also recruits for military service" second
the farms 'would encourage some at least of the nomads t~ adopt ~
settled agncultural life'; and third, the 'fron tier farmers could be
encouraged to act as part-time frontier guards'.
15 1 Th '. f
. . e quotatlOn l~ rom Le Bohec, La Troisieme Legion Auguste, 57 8. The idea
that thejos~atUl~ was mten?ed to war.d off raiders from the desert goes back at least
to J. p~ey, FOUllIes su~ le lImes romam de Numidie (dans la region de Bordj Saada
ep fevne,r.-~ars, ~938) '. CRAI C.1~38), 357--9; it is neatly summarized by Leschi:
E.!udes ~ eplg:aphle, 7 1 : en Numldle, des mesures militaires ont, des le debut du IIe
sJec~e,. mt~rdl! ~ux Nomades, pa: u~e solide barriere, Je libre acces des terres
cultlvables. Slmllarly, Baradez mamtamed (Fossatum Ajdcae 148) that the S .
Bent ~l ~ras,s section of thc jossatum was meant to block acc~ss ro~tes to 'zon~~u~:
co!olllsatlOn; c~ Fentres~, Numidia, 1 r I: 'the Gemellae segment circumscribes all
the usable lan~ m the ~egl?n ~f the Oued Djedi' (my italics); Williams, Reach, 122.
See als~ Mattmgly, Tnpohtama, 37, and the excellent discussion in B D Shaw 'The
Camel m ,Roman .North, Africa and thc Sahara: History, Biology a~d 'thc H~lman
Economy , Bulletm de l Institut Fondamental d'Ajrique Noire 41 ('979) 663- 2
esp. at 674-5. ' , 7 1,
I~2 M~xfield, 'Front~ers', 343 (my italics). See also Raven, Rome in Ajrica, 91 ;
~pcl~el, Roman Arm~ , 406 (the fossa near Sala in south~west Mauretania Tin i~
tana I?rot~cte? the agnculturally valuable territory'). For a similar argument abogut
th~5rr~poht~m~n fronti~r, see .Re~uffat, 'La Frontiere romaine', esp. at 233.
. ,H:'ldnamc FrontIer Pohcy, esp. at 29-32. So, too, Daniels 'The Frontiers'
Afnca , 1.. 250: 'Hadrianic fron tier policy had been to encourage 'economic devel~
opment nght
U . V' to the. actual limes.' See also G . A . "\Hebster
up H'
" N 0 t e on N ew
, rt
lsco~e:les ~t .1roC~01~~1 ~Wroxeter) Which May Have a Bearing on Hadrian's
FrontleI POb~Y.lll Bntam, m Hanson and Keppie (eds.), Roman Frontier Studies
/979, 295; WIihams, Reach, p. xix.
The Roman Frontier-Zone 61
The argument fails for at least two reasons. It assurnes that the
imperial government set out to populate the frontier-zone witb
Italian immigrant-farmers; in Birley's words, 'tbe frontier region
reeeived a eonsiderable eivilian population, and in the Biskra
region that cannot weil have come about without direct support
from the emperor, as an aet of imperial poliey'. But there is no
evidence of extensive immigration to the Algerian frontier-zone;
the expansion of (mostly olive) cultivation in the area that was
aehieved in the period of tbe Roman oecupation is very likely to
have been a native accomplishment. 154 The argument also assurnes
what is demonstrably incorrect, that the frontier-line eoincided
with the intersection of agriculture and pastoralism: the forts
and barriers of the frontier in fact cut across the traditionallands
of some tribes (like the Musulamii) who praetised seasonal trans-
humance. '55 There are no grounds far thinking that the fron tier
was ever intended to be a barrier between sedentarists and
nomads. I56
A somewhat different explanation has been advanced by Brent
Shaw, who argues that the several segments of the fossatum were
meant to protect settlements, not, as has generally been thought, in
the northern plains, but in the exploitable 'belts' of land that run
along tbe bases of the Saharan mountain massifs; their purpose, be
suggests, was to ward off 'low intensity' threats from pastoral
tribesmen, neighbouring communities or 'the inhabitants of the
mountains above'. r 57 I suggested earlier that the Roman army is
unlikely to have built linear barriers whose sole (ar even main)
purpose was to protect civilian settlements and that tbe fossatum
was probably not intended to protect agriculturalists (or, for that
matter, anyone else) against raiding activities: the Romans must
have known that determined bandits could easily cross it. And
while it is entirely possible that the Hodna mountain segment of
the fossatum was designed, at least in part, to ward off threats (of
154 See B. D. Shaw, 'Water and Society in the Ancient Maghrib: Technology,
Property and Development', Ant. afr. 20 (1984), T21-73, esp. at 163.
155 See Le Bohec, La TroisiCme Legion Auguste, 541; Whittaker, 'Trade and
Frontiers' , 112; 'Supplying thc System', 66.
15 6 P. Leveau, 'Le Pastoralisme dans I'Afrique antique', in C. R. Whittaker (ed.),
Pastoral Economies in Classical Antiquity (Cambridge, 1988), 181: 'Il est demontre
qu'au Maghreb le limes n'est pas la limite entre vies nomade et sedentai~e.' The
point is made also in Shaw, 'Camel in North Africa', 676; 'Fear and Loathmg', 41.
J 57 'F ear and Loathing', 42.
 62 The Roman Frontier-Zone
some kind) from the upland tribes, the same eannot be said of its
other seetions. The Seguia Bent el Krass segment south of Gemellae,
for example, was located, not between the mountains (in this case,
the Zab and Aures) and the communities that ean be assumed to
have dotted their southern reaches, but beyond (that is, south of)
the agrieulturally valuable land; it must have been intended there-
fore to deal, in some fashion, with people headed north from the
area of the Chott Melrihr (see Fig. 2.6).
My own position is that the Roman fron tier in Algeria and its
various fortifications had two main funetions: to provide for the
security of the soldiers themselves; and to enable the army, and, by
extension, the imperial government, to tax the products of pastor-
alism. There is a tendency in modern thinking about the north
African fron tier-system, observable in virtually every piece of ser-
ious writing about it that has been published in the last twenty
years or so, including Edward Luttwak's influential study of
Roman imperial strategy (The Grand Strategy of the Roman
Empire fi'om the First Century A. D. to the Third), to assurne that
the various parts of the system were designed to accomplish a
single purpose. So when it is concluded, as it inevitably must be,
that the fossatum cannot have been intended to function as a
military barrier, the entire system is treated as having had no
defensive application. The historian is led inexorably to the rather
ludicrous position of arguing (usually implieitly) that the Roman
army built forts which had no real military purpose.
The obvious solution is to admit that the different parts of the
system are likely to have had different funetions. '58 The network of
Roman military roads and forts must neeessarily have been con-
structed, at least in part, to counter a perceived threat to the seeurity
ofthe soldiers or oftheir lines of eommunication and resupply. '59 In
this eontext, it matters little whether the threat was real or imagined.
What seems to be shaping up as today's orthodoxy, in whieh it is
denied implicitly that Roman military strategies on the desert fron-
tiers may have been determined by perceptions whieh we know to be
false, is methodologically indefensible.
We know now that Saharan nomads eould never have ser-
iously threatened the Roman order; Louis Lesehi's vision of
IS8 Cf. Wheeler, 'Methodological Limits', IO-II.
'59 cr. Isaac, Limits 0/ Empire, 69, on the eastern frootier.
The Roman Frontier-Zone
camel-mounted raiders swooping out of the desert to pillage
'Roman' farms and settlements is, in hindsight, folly. ,60 But we
have no reason to believe that the Roman army knew that. It was
remarked earlier that the Romans appear to have had very httle
information of any kind about the peoples who lived beyond the
frontiers. Only two Roman expeditions (thos.e of Gams Suetomus
Paulinus and Gnaeus Hosidius Geta) are sald ever to have pene-
trated the western Sahara, and then only, in asense, madvertently
(in pursuit of rebels). So while it may be agreed that the Saharan
nomads posed no serious threat to the frontier in the penod before
ab ou t AD 250 , ,6, it eannot also be said that the Romans dld not
, 162
consider them to be 'a major seeurity problem. .
It is not unlikely either that some parts of the front~er-system
were intended to counter internal threats to the army s secunty
(again, in this context, it matters little whether they were genu-
ine).'6 3 Perhaps the clearest example is the TraJame constructlOn
of roads and forts south and west of the Aures mountams, whleh,
it is generally agreed, was designed to encircle them. We ~annot
expeet ever to know whether the Aures really were, as Benabou
believes a continuing source of resistance to Roman rule. I 64 What
counts ;s that the Roman army evidently thought that they were.
Other parts of the frontier-system, and most espeemlly lls hnear
barriers were obviously not intended to serve a defensIve purpose.
It may be agreed that thefossatum was designed instead to 'contral
and direct transhumant tribes'. J65 The nature of the barner 18
160 Leschi: Etudes d'epigraphie, 65-74. "
16! So Fentress, Numidia, 117; Isaac, Limits oj EmpIre, 4~4.
162 The quotation is from Is~ac, L.imfts, 01 Empire, 99 (smd of eastem nomads).
See also Wheeler, 'Methodologlcal Llmtts , 33, 35· . .. th
163 Cf. Shaw, 'Soldiers and Society', 138: the Roman ~mear b~rne~s 1ll nor
Africa may have been 'multifunctional', directed, at least 1ll part, to hlghly local,
and internal threats'. . Wh'te 'Geo-
4 Benabou: La Resistance africaine, 73· So, too, A. N. Sherwm~
16
h1'B b
.
graphical Factors m Roman Al'
germ,• JRS 34 (1944) 5'
, .','as a whoie
..teer
. t er d
o ulation of the Aures remained undisturbed and h?sttle " The no.tlO';11S reJe~ e
byPMorizot 'Vues nouvelles', esp. at 3II, 337; cf. Thebert, Romall1S~tlOnhe~ ~er~­
manisation" 75. See also Shaw, 'Autonomy and Tribute', 69, suggestmg t a ec-
nological li~itations impeded the Romans' ability to 'impose an adequate control
over mountain frontiers' . b 6' F
16 5 Whittaker, Frontiers, 147-8; see also Brett and Fentress, Ber ers, 7, en-
tress Numidia II2' Williams, Reach, IS0. It has been suggested, too, th~t the
distribution of Rom'an troops across the Algerian fron tier-zone, at. Lambaesls and
at many smaller legionary garrisons, was designed to control tnbal movements
 The Roman Frontier-Zone The Roman Frontier-Zone 65

barriers to canalize the movements of pastoralists and their ani-

Fossatum mals. I argued earlier that it is doubtful, for a number of reasons,
~ Transhumants that its purpose was to regulate the interrelationship of agriculture
Zarai and seasonal transhumance. It may be said again that tbe army is
Diana unlikely to have constructed and maintained the fossatum other
Veteranorum than in the expectation that in so doing it would acquire some
material or other advantage.
What I suspect to be the fossatum's real purpose was remarked
long ago, in a somewhat elliptical way, by Birley, who, after
arguing at length that the fron tier was intended to foster agricul-
Lambaesis tural settlement, went on to say that the 'Iife of a Hadrianic
frontier centred more in customs and passport control'. r67 There
are, it seems to me, several reasons for thinking that the different
segments of the fossatum were designed to direct transhumant
traffk towards places where it could more readily be taxed. ,68
Aures Mts. Pastoralists heading north from the region of the Chott Melrihr
toward the Tell are likely to have run into the Seguia Bent el Krass
segment of the fossatum; any who tried to get around it to the west
Vescera will have run up against the southern reaches ofthe Zab mountains,
Zab Mts. Thabudeos
where they will have been turned north-east towards Gemellae and
Vescera; likewise, those who went east to get around it will then
~ Gernellae have had little choice but to proceed north toward Thabudeos and

~ , Vescera (see Fig. 2.6).'69

In much the same way, transhumants from the Zab mountains
headed towards the EI Kantara pass of the Aur!:s will have been
stopped short by the Mesarfelta-Thubunae section of the fossa-
Fig. 2.6. Linear barriers and transhumant routes in the jrontier-zone tum, and diverted, either north towards Thubunae or south to
Mesarfelta. A similar purpose may be attributed to the sec!ion
d:cisive~while people could have traversed the ditch and wall at of the fossatum that appears almost to have encircled the eastern
t e Segma Bent el Krass generally without much difficulty, it end of the Hodna mountains. The many pastoralists who wintered
would ,~~ve been alm ost Impossible to get, say, a flock of sheep their animals in the grazing land of the Hodna basin and who were
across. The real questIOn then is why the Roman army built accustomed, if modern transhumant routes are a reliable guide, to
drive their animals north in the early summer over the eastern
('deplacements de populations') . 11 1
'Recherches a Lambe'se' , especra y a ong transhumant routes: lanon reaches.of the Hodna mountains, will have been turned first east
, 199· '
166 Th by the Hodna section of the fossatum, and then north or north-
the sameef~~~:iz;;:;e ~:e~~o~~~;~P~lit:i~ are ~nderst~od to have performed mueh
controls on shepherds and herds!e~ Wh~t~a~~t. Fro~tlers, 81: they were 'internal
'Hadrianic Frontier Policy', 33-
Ws as made first (as far as I ean tell) in Trouss;: ~;~~:rche~~verr1 them'. The point
r67
168 Cf. Raven, Rome in Ajrica, 77-8; WeHs, 'Problems of Desert Frontiers' ,47 8:
ee also P. Trousset 'Note sur un t d' '.,. ur e l»leS, esp_ at 140-1. the soldiers stationed in north Africa 'might be intended primarily to carry out
(19 84), 18 3-9 8 - Ma~tingly T. - l"iP~ ouvrage !meaIre du limes d'Afrique', BCTH
~7, 94. ' , npo lama, 79; Mattmgly and Iones, 'A. New Clausura', administrative or tax functions, especially in the fertile aases'.
169 For this and what follows, see especially Fentress, Numidia, II2, 1 67_
 66 The Roman Frontier-Zone
east by the Aur"s, with the resul! that they will have had to pass
through or near Zarai, Diana Veteranorum, or Lambaesis,
The purpose of these various arrangements is indieated I
beheve, by two early third-eentury AD inseriptions whieh ree~rd
the eustoms dutJes that soldiers were expeeted to eolleet on goods
transported across the frontier-zone. 170 One, from the civilian set-
tlement at La~baesis (AE (1914), 234), is ineomplete; enough
survlves to mdICate that soldiers from the legionary fortress were
used to supervIse the eolleetion of tolls and tariffs in aeeordanee
wlth a lex, portar;;, '7' The other (eIL 8, 4508), unearthed at Zarai
m 18 58, IS more instruetive, Tariffs are listed for a number of
prod~ets, many of which may be assoeiated with pastoralism:
they mclude woollen goods (tunies, blankets, and cloaks), a variety
of ammals (horses, mules, asses, eows, bulls, pigs, sheep, goats),
leatber, hldes (of various kinds), dates, figs, peas, nuts, 'glue, resin
plteh, wme, fish-sauee (garum), sponges, and slaves, '7' We migh;
suppose that there were similar tariff-lists at Vescera Mesarfelta
Thubu?ae, and Diana Veteranorum. That they ha~e not bee~
found IS not, of course, proof that they never existed (there has
been no systematie excavation at any of the sites).
My eoncluslOn then is that the Romans built the fossatum
mamly to faeIlltate the eolleetion of taxes on goods brought north
towards the Tell. '73 The real objeetive, I imagine, was to enable the
army to help pay for Itself. At various times, the Romans have been
smd to have had, other purpose,s i~ the frontier-zone. They may be
s~m;nanze~ as. RomamzatlOll, urbamzation', and 'sedentariza-
lIon . My vlew IS that there is no real evidenee for any of them.
Romanization
It was Benabou's understanding that the Romans set out
aggresslvely .to Impose their values and eustoms on their north
Afflean sub]eets, msplred, it seems, by a belief iu their eultural
:~~ See especial13:' Darmon, 'Note sur le tarif',
'7 2~~e also Davle~, Service, 61-2; Janon, 'Remarques sur la frontiere' 483
, . MacKendnc~, North African Stones, 248-9; Raven Rame in Afric~ 78 -
Sh~'i' !ear .a~~ Loathmg', 42-3. The tariff-rates were gener~lIy less than 3%.'
. . f: Wtllrams, Reaeh, ISO. Fentress, Numidia, 116, suggests that Castellum
Dlmmldl may have beel~ built to control and tax a trade route from the Niger rive
to .Laghou~t. But t~ere IS, as she admits, no evidence to show that such a route eve~
eXlsted. It IS more hkely, I think, that the fort was intended to be a forward sur
post.
The Roman Frontier-Zone
superiority.'74 Graham Webster takes a more eharitable view of
Roman motives, but is no less eonvineed of their objeetive: the
'frontier provineials must be made to feel they were Roman
and so in developing their way of life in peaee and prosperity, to
defend it against the barbarians'. '75 Webster at least understood
the Romans to be self-interested. Not so Eadie, who eriticizes
'revisionists' for portraying the Romans as 'insatiable plunderers';
their error, he believes, is a produet of their failure adequately to
appreciate 'the Romau attempt, fundamental to the development of
their fron tier polieies, to establish the regions on both sides of the
border as zones of acculturation' .I 7 6
This relentlessly optimistie view of the aims and effects of the
Roman occupation overlooks an important truth: there is no evi-
dence, of any kind, for a deliberate poliey of Romanization in north
Africa or in any of the other provinees of the empire. '77 Those who
would have us believe otherwise ean point to only a single passage
in the whole of Latin literature, Tacitus' now-famous description
(Agr. 21) of his father-in-Iaw's programme, when governor of Brit-
ain in the late 70S AD, that arranged for the sons of the leading
Britons to be taught Latin, and that eneouraged the natives to
build 'temples, fora, houses','78 Agrieola's programme cannot be
174 'Les Romains', esp. at 87. 175 'Note on New Discoveries', 295.
17 6 'Civitates and Clients', 57; cf. 'Peripheral Vision in Roman History', in 1.
D'Arms and 1. Eadie (eds.), Ancient and Modern Essays in Honor 0/ Gerald F. Else
(Ann Arbor, 1977), 222: 'the objective of Roman fron tier policy' was 'the accultura-
tion of barbarian groups'. See also Birley, 'Roman Frontiers' , 18, who thinks that a
policy of Romanization may be implied in Aristides' speech 'To Rome'. I do not
understand why Dyson, ereation, 116, describes road-building as an 'instrument' of
Romanization.
I77 See D. B. Saddington, 'The Parameters of Romanization', in Maxfield and
Dobson (eds.), Roman Frontier Studies I989, .413. Cf. Drummond and Nelson,
Western Frontiers, I28: the Roman imperial government had 'little intention' of
making a 'major effort' to acculturate the fron tier populations; P. 0rsted, Roman
Imperial Economy and Romanization (Copenhagen, 1985), 20: Romanization was a
'deliberate policy with economic goals'. Millett, Romanization oi Britain, 10I,
concludes that in some parts of Britain, Romanization was 'stimulated by passive
encouragement'. But cf. Millett, 'Romanization: Historical Issues and Archaeolo-
gical Interpretations', 37, where he suggests that we must question 'any idea that
Rome had any real policy of Romanizing its conquered territories beyond ensuring
' an effident administration'.
17 8 cr. D. 1. Breeze, 'Demand and Supply on the Northern Frontier', in R. Miket
and C. Burgess (eds.), Between and beyond the Walls: Essays on the Prehistory and
History oi North Britain in Honour oi George Jobey (Edinburgh, 1984), 282: 'it
would be hard to find a parallel in ancient literature for the encouragement by
vey central government of Romanisation in this way'.
 68 The Roman Frontier-Zone
shown ever to have been duplicated in Britain or in any of the
other provinces. '79 There was, as Ramsay MacMullen has put it,
no policy behind Agrieola, 'not even a tendency'.' 80
An alternative model is proposed: Romanization, it is said, 'was
never really a policy of raising native peoples to the level of Roman
culture'; the Romans tried only 'to convert the loeal elites to a
Roman way of life'. ,8, The mistake here is to confuse a Roman
strategy that aimed at co-opting local elites to share in the costs
and responsibilities of administering tbe empire-wh at Peter
Garnsey has called a 'traditional' programme of 'building up a
network of families, groups and communities with vested interests
in the prolongation of Roman rule'-with a policy of acculturating
provincial aristocracies. ,82 The point has been made forcibly of
Roman policy in Tripolitania: the Romans' real purpose there, in
David Mattingly's view, was 'to create stable conditions of local
government'; at a 'far lower level', the Romans hoped that tribaI
elites within and beyond the frentier might be 'reconciled' to
Roman authority; the goal was not to turn Libyans into Romans,
but to persuade potential enemies 'to identify just a little with
Roman civilisation,.'83 In north Africa at least, the process of
179 See also Th6bert, 'Romanisation et deromanisation', 72 .
180 MacMullen: Changes, 57; cf. Hanson, 'Nature and Function', 59.
181 Drummond and Nelson, Western Frontiers, I91 (it may be remarked, in
passing, that J know of DO objective criteria according to which the 'level' of ancient
cultures can be ral1ked). On Roman efforts to acculturate Ioeal eIites, see also N. J.
Higham, 'Roman and Native in England North of the Tees: Acculturation and its
Limitations', in Barrett, Fitzpatrick, and Macillnes (eds.), Barbarians and Romans,
154: 'thc process of acculturation', a 'conscious cornponent of Roman imperial
policy in Western Europe', was 'aimed at the aristocracy'. Eadie, 'Civitates and
Clients', 65, takes it a step further: the Romans attempted to acculturate both dient
kinfs and native civitates.
1 2 Garnsey: 'Rome's African Empire', 235. See also D. Braund, 'Ideology, Sub-
sidies and Trade: The King on the Northern Frontier Revisited', in Barrett, Fitzpa-
trick, and Macinnes (eds.), Barbarians and Romans, I4-26; P. A. Brunt, 'The
Romanization of the Local Ruling Classes in the Roman Empire', in D. M. Pippidi
(ed.), Assimilation et resistance a la culture gr(!Co-romaine dans le monde ancien:
Travaux de Vle Congres International d'Etudes Classiques (Madrid, Sept. 1974)
(Bucharest and Paris, 1976), 161-73. A case could be made, too, pace MacKendrick,
North African Stones, 148, that Roman efforts to encourage emperor-worship in the
provinces were designed, not so much to facilitate the acculturation of local elites,
as to disseminate a highly visible symbol of the political and economic stake which
th~r had a~quire~ i? the continuatiol~ of Roman rule. .. . .
3 Mattmgly: LIbyans and the Limes: Culture and SocIety 111 Roman Tnpoh-
tania', Ant. afr. 23 (1987), 80; cf. 72: 'the Romans were not trying to enforce a
complete cultural complex on their subject peoples'; see also Tripolitania, 166.
The Roman Frontier-Zone 69
acculturation was, as Broughton remarked long ago, 'too gradual
' l pressure ,,84
to have been the result of 0 ffi CJa . ..
It has sometimes been suggested, too, that tbe Roman Impenal
government planted colonies of veteran~ in the frontier-zo~es to
'assist' in their Romanizatic)fl. 18 5 Accordmg to Fentress, the m~r~­
duction of veterans into the settlements (vici) of soutbern NumldJa
that had been formed out of pre-existing communities will have
'facilitated their organisation and romanisation,.'86 The idea
cannot be shown to be [alse-no Roman source discusses the
purpose of tbe veteran colonies. But it must surely be the case
that their main function was to provide the veterans with land. ,87
Even if they were intended to serve also as 'cent~rs of Romamza-
tion' tbey are unlikely to have been very effeclIve-most of the
vete:ans will have been of provincial origin.
188
Urbanization
It is no more likely that the Roman imperial government tried to
urbanize the frontier-zones. It has been said of north Afnea that
'those of the natives who were still nomads or semi-nom~ds' we~e
'encouraged to develop urban communities'. 18 9 How mlght thiS
have been accomplished? Are we to imagine that the emperors
184 Romanization 01 Africa, 113; see also 134, 141 ('Romanization was not a
Roman policy'), 224· . 'Ad .. _
18 5 Drummond and Nelson, Western Frontzers, 140; see also ~an~on, mmlS
tration' 53-4 (they were 'models of the Roman lifestyle for the mdlge~o.us popu-
. " 186 Numzdw, I40-I.
latIOn ). .
187 Cf. Breeze, 'Demand and Supply', 282: 'the settlement of veterallS Ir: m.any
regions in frontier areas, freque~tly cit~d as support f?r a policy of RomamsatlOn,
is of course connected with entlrely dIfferent factors. . .
' 188 'Cent~rs of Romanization': P. MacKendrick, 'Roman CololllzatIOn an~ t~e
Frontier Hypothesis' , in W. D. Wyman and C. B. K.roe?er (eds.), . The Frontzer 111
Perspective (Madison, 1957), 13. Broughton, RomanzzatlOn .01 ~/flca, 117, argue~
that the function of the veteran colonies was to 'w~tch the lOdlge~ous tnbesmen ,
the were not, he suggested, 'an attempt to Romatllze tbe country. . . .
1~9 1. A. Ilevbare, 'Some Aspects of Sodat Change in ,North Afnca l,n. PUUlC
and Roman Times', Museum Alricum, 2 (1973), 32. Benabou, La ,:eszst~nce
alricaine, 30, takes a similar position. See also J. R. F. Blo~m~rs~ RelatIOns
between Romans and Natives: Concepts of Comparatlve StudIes, 111 ~axfiel~
and Dobson (eds.), Roman Frontier Studies 1989, 453 (on Lower Gelma~y)~
Drummond and Nelson, Western Frontiers, 127; ·A. T. Fear, Rome and Baeyca.
Urbanization in Southern Spain c. 50 BC-AD 150 (Oxford, 1996), 15; MIllett,
'Romanization: Historical Issues', 39·
 70 The Roman Frontier-Zone
considered it to be their responsibility to pay for the construction
of towns m the frontier-zone?19 0
. It cannot be denied that the number of urbanized settlements
mcreased, perhaps dramatically, over much of north Africa in the
penod of R~!?an ?ccupation, and especially, it seems, in the second
century AD. It IS not, however, a development that the Romans
can. be shown to have directed. '9' The army of the post-conquest
penod may be sald to have supplied the necessary conditions, of
secunty and repose'.'93 But the urbanization that is attested in north
Afnca was pnmanly a native development, originating in and
fuelled by competItlOn among Romanizing local elites. '94
Sedentarization
It has sometimes also been said that the Romans had a policy of
sedentanzmg the north Afncan pastoralists. According to Paul
Mac~endnck, the 'central aim' of the Roman military system
was to make the nomads sedentary'. '95 J. A. Ilevbare was
equally certam that the nomads were 'systematically compelled
::~ The point is made in Isaac, Limits of Empire, at 334.
~'. P Duncan-Jones, 'Wealth and Munificence in Roman Africa' Pa ers of
thed Bnflsh
H' School at Rome" JI (196J) 170',B. D. Sh aw, 'CI'lmate E'nVlronment
. 'P
an lstory: The Case of Rom~n North Africa', in T. M. L. 'Wigley, M. 1.
I~g.r.a~, and G. Farmer (eds.), ~hmate and History: Studies in Past Climates and
~ eU ,~pact on Man (Cambndge, 1981), 392; Sherwin-White, 'Geographieal
ae t ors, 9.
R 19 See aI so Bro~ghton, Romanization of Africa, who repeatedly denies that the
2
o~ans had a pO!ley.of urbanization in north Afriea, e.g. at 226: 'urbanization
pal tlcularly urballizatlon on the Roman model was not a Ro " '1
poliey'; cf. 12 7,224. , m a n or an lmpena
. 193.This is pr~su~ably what .Strabo meant when he declared (3. 3. 8) that
Tlbenus, by statlOnmg three l~glOns in Cantabria, n:tade the region 'fit not onl
f~r p~ace but even for urban Irfe'. R. I. Lawless, 'L'Evolution du eu Plement d~
I habitat et des paysages agraires du Maghreb', Anna/es de geogra%hie 81 (I ' 2
457, 459, ma~es a similar point about 3rd-cent. AD frontier-zone s;ttlem 91)'
~estern.Al~en.a; cf. 1. Mertens, 'The Military Origins of Some Roman Settle:en~~
111 B~lgl~~, In B. Hartley and 1. S. Wacher (eds.), Rome and her Northern
~;eov~~e~. l}:pzrs Pr~sented to Sheppard Frere in Honour of his Retirement from
GI wr 0) t e Archaeol?gy of the Roman Empire, University of Oxford 19 83
( ouc~ster, 1983), 164; Millett, Romanization ofBritain 99' see also Isaac 'L' "
of EmpIre, 420. ' , , Iml s
194 C R Wh'
.. lttak
er, ' Do Theones
. of the Ancient City Matter?', in T. 1. Cornell
an~r· K: Lom~s (eds.), Urban Society in Roman Italy (London, 1993), 8.
. dNOlth A[r.Ic~n Stones; 329. S~e also Benabou, La Resistance africaine, 7I: the
noma s were vlctImes de I enterpnse romaine'.
The Roman Frontier-Zone 71
to settled life'. '9 6 The Jossatum itself, it has been suggested, was
designed to blockade the pastoralist tribes and thereby to force
them to adopt an agricultural economy. '97 The Romans' purpose
in all this is generally understood to have been either the exten-
sion of cereal culture and arboriculture, or the expropriation of
tribaI lands. '9 8
lt is undeniable that the characterization of nomadic life which
survives in Roman sources is uniformly hostile. 199 The real issue is
whether the Roman army, as a matter of policy, set out system-
atically to confine the pastoralists to designated regions (what
Benabou, among others, calls a policy of 'cantonnement,).200
The bulk of the evidence that is thought to demonstrate the
existence of such a policy is epigraphic: boundary-stones (cippi)
that the Romans erected in many parts of north Africa, usually, it
seems, in the course of cadastrating tribai lands. 2or The earliest
surviving examples are from the area that lies south of the
Tacapae-Ammaedara road and north of the Chott el Fedjedj;
dated to the late 20S AD (perhaps to 29-30), they were set up
by the Third Augustan legion on the orders of C. Vibius Marsus.
One (eIL 8. 22786) is inscribed with the letters NYBG, which
undoubtedly refer to the Nybgenii, and which are often
interpreted to mean that the tribe had been assigned land (a
19 6 'Aspects of SociaJ Change', 32. So, too, M. K. Abdelalim, 'Libyan Nation-
alism and Foreign Rule in Graeco-Roman Times', in Libya Antiqua: Report and
Papers ofthe Symposium Organized by UNESCO in Paris, 18 to 19 January, 19 84
(Paris, 1986), 156 (on Libyan nomads).
197 Leschi, Etudesd' epigraphie, 65~74; Rachet, Rome el fes Berberes, 79-80, 164-73.
19 8 Extension of cereal culture: Rachet, Rome et fes Berberes, 68. Expropriation
of tri baI lands: Benabou, La Resistance africaine, 43 1.
199 See B. D. Shaw, 'Eaters of Flesh, Drinkers of Milk: The Ancient Mediterra-
nean Ideology ofthe Pastoral Nomad', Ancient Society, 13-14 (1982~3), 5-31, esp.
at 5. There are excellent discussions ofnomadism in 1. Despois, Le Hodna (Algerie)
(Paris, 1953),265-91, and in S. Lancel, 'Suburbures et Nicives: Une inscription de
Tigisis', Libyca, 3 (1955), 289-98. See also P. M. E. Lorcin, Imperial Identities:
Stereotyping, Prejudice and Race in Colonial Algeria (London and New York, 1995),
51, I90, on French colonialist attitudes to Algerian (semi-)nomadism.
200 Benabou: La Resistance africaine, 429-45; so, too, Brett and Fentress, Ber-
bers, 62; Leveau, 'Le Pastoralisme', 178. Broughton, Romanization of Africa, JI6,
maintains that the army forced the (semi-)nomadic tribes to 'settle on limited
territories' (cf. 96, 121, 126-7)·
20) The arrangements are described in A. Marcone, 'Note sulla sedentarizza-
zione forzata delle tribli nomadi in Africa aBa .luce di a1cune iscrizioni', in A.
Mastino (ed.), DAfrica Romana: Auf dei IX convegno di studio, Sassari (Sassari,
1992), 104- 14.
 The Roman Frontier-Zone
'reserve') in the region. 202 Inscriptions of the early 80S are said to
show that similar arrangements were made in the territory of the
Suppenses and Vofricenses, and of the Suburbures and Nicives
(in the plain of Ain Abid, south-east of Cirta). 203 Trajanic
boundary-stones (CIL 8. 22787-8) set up in the territory of the
Nybgenii are taken to indicate that they were henceforth 'limited
to the zone of the Chott el Fedjedj'.204 A number of inscriptions
record cadastration in the territory of the Musulamii in the time
of Trajan and Hadrian. 205 Other Trajanic boundary-stones (ILS
9380-1, AD II6-17:ji.nes adsignati) are understood to be evidence
that the Suburbures, or apart of the tribe, were assigned land in
the region of tbe Chott Beida. An unidentified 'Numidian' tribe
(gens Numidarum) is attested near Bordj Bou Arreridj (south-
west of Sitifis) in the time of Hadrian.,06 A Severan-era inscrip-
tion (BCTH (1917), 342-3) seems to show that at least some of
the Suburbures (res publica gentis Suburburum colonorum) had
taken up residence in the region of Azziz ben-Tellis' the Romans
it has been suggested, were now forcing the transhumant tribe~
south and west towards the desert.'07 Another Severan inscription
(AE (1946), 38), from the region south of the Chott el Hodna
records that a three-man commission, acting on the orders of th~
legionary commander, assigned 'fields and pastures and springs'
(agri et pascua et fontes) to persons who are unidentified but
likely to have been indigenous pastoralists. '08 ' closely to tbe model of 'imperialistic' frontier policy that has been
. What is missing in the various epigraphic reports is any indica-
tlOn of the Romans' purpose. It is necessary to insist that the
~o~ s~~ e.g. R.aven, Rome in Africa, 88; cf. Fentress, Numidia, 68, 73. For the
Nybgenn, see Phny, HN 4. 27.
3 S~ppenses and Vofricenses: AE (1942-3),. 35: termini repositi inter Suppenses
20
et V~fi.·lc:nses; ~uburbu~es a~d Nicives: AE (1957), 175; AE (1969-70), 669: agri
pU~i!CZ Czrt(enszum) .a~slgnatl. See al~o Garnsey, 'Rome's African Empire', 232.
5 Fentress, Numldza, 73; cf. Mattmgly, Tripolitania, 77.
20
~1 AD 100/3: CIL 8. 10667 (:::: 16692 :::: ILS 5959). In 104/S: CIL 8. 467 6 (=
2807] - ILS 5958a); AE (r907), r9, 2r; AE (r92J), 26. In r r6: CIL 8. 2807]b (= ILS
595 b); AE (I9~7), 20; ILAIg I. 2939bis. In 138: CIL 8. 270 (= 11451 :::: 23 246). See
8
also the table 1ll Fentress, Numidia, 74-5, and the discussion in Benabou La
Resistance africaine, 438. ' ~I~ See also WeHs, 'Problems of Desert Frontlers, 478; cf. <?arnsey, RO,me s
~06 CIL 8. 8813-14; see also Broughton, Romanization of Africa, 12 5.
~07 See Garnsey, 'Rome's Afriean Empire', 23 2.
208 See also Trousset, 'De la montagne au desert', 100-1. The evidence for 'land
delimitation' in Trip~1itania is ~xami~ed in Mattingly, 'Libyans and the Limes', 83,
who conc1udes that It was deslgned to transfer the title of some lands from the
whole tribe to individuals'.
The Roman Frontier-Zone 73
inscriptions do not demonstrate that the pastoralist tribes were
compelled to become farmers or confine d to "1ll1enOr C • , I
and s. '°9
They show only that parts of some tribes were (or became)
sedentary in agriculturally productive regions; the bo~~~ary­
stones may have been intended to define thelr terntones. . The
cadastration of tribally controlled regions was probably deslgned
either to prepare the land for immigrant settlement or, more
likely in my view, to establish the tax liability of the tnbal
seden tarists. 21 I
The Romans in fact had a number of good reasons not 10
sedentarize the semi-nomads: the products which they brought
north (leather, wo ollen goods, etc.) could be taxed; their labour
was needed in the Tell at the time of the harvest; and thelr ammals,
as was noted earlier, helped to fertilize the fields planted for cereal
culture. 212
CONCLUSION: ROMAN OBJECTIVES IN THE FRONTIER-ZONE
Putting even the best face on the Romans' actions in the region: it
is hard not to conclude, with Broughton, that thelr only ldenllfi-
able policy in the fron tier-zone is one of 'exploitation'. 213 The
characteristics of Roman rule in Algeria would seem to conform
developed by B. Barte!: little immigration mto the fronller-zone;
hardly any change in native economic structures; hmlted mterac-
214
tion between the intrusive and indigenous cultures.
;209 See Graf, JRS 82 (1992), 278; Lancel, 'Suburbures et Nieives'; Shaw, 'Fear
and Loathing', 44. Garnsey, 'Rome's African Empire', 232, also eonc1udes that
'there was no question of suppressing nomadism'; ~u~ cf.. ?33, where he dec1a~es
that 'the participation of at least the trib~lleadershlp 111 eitles su~h as Thuburs.~eu
Numidarurn Gigthis and Turris Tamellem ... suggests that a poltcy 0/ sedentarz",a-
tion had achieved some success' (my italies). .,
;210 See especially Berthier, 'Nieibes et Suburbures'; Whlttaker, Lan~ and
Labour', 345-6. 21 I Wh~tta~er, 'Land and Labo~r , 34~.
African Empire', 232-3; P. Leveau, 'Oceupation d"';l sol, geosystemes et sys~eme,s
sociaux: Rome et ses ennemis des montagnes et du desert dans Je Maghreb antJque,
Ann. ESC 41 (1986), 1345-58; Raven, Rome in Ajrica, 77·
U3 Broughton: Romanization of Africa, esp. atA4·
;214 Bartel: 'Colonialism and Cultural Responses: Problems Rela~ed to. Ro~an
Provineial Analysis', WAreh. I2/1 (1980), II-26; 'ComparatJve Hlstoncal
 74 The Roman Frontier-Zone
There IS no basis for believing that the Romans' purpose in
Algena was to defend the loeal population or the land it eultivated
against ineursions by (semi-)nomadic peoples who lived beyond
the frontIer. There IS no reason to suppose either that they con-
sldered It to be their duty to manage the interdependence of
agnculture and pastoralism. And there is no good evidenee of
any kInd to show that their intention was to make north Afrieans
lilto Romans, or pastorahsts wto sedentarists.
My own view is that the imperial government's real purpose in
Algena was to generate tax revenue, and to provide for the seeurity
of ItS agents (soldlers and administrators). The arrangement of
forts, roads, and lInear barriers in the frontier-zone strongly sug-
gests that the soldiers stationed there were meant to function
~nmanlY as an army of oceupation. In this sense, at least, it may
e agreed wI~h MacMullen that the Romans' main objective was to
monopohze the force of arms'. 2IS
Archaeology and Archaeological Theor ' in S L D . An Introduction to Anthropological Strategies Jor the Study of Sociocultural Change
8)
in the Archaeology 0/ Colonialism (Oxf;rd 19 /son ~ed')'l Comparallve Studies
;:u~~~I:I~:!::a~~~~:ssY: (~~~~~~~oation' in t~; E~~l; ;~':aS: ~~~ie~~~;f~~
Zl5 MacMulJen: Changes, 57.
3
Measuring Romanization
Historians and archaeologists have long debated the nature and
extent of the acculturation of the Roman provinces, perhaps
more precisely, since the movement is usually understood to
have been largely one-way, the Romanization of their indigenous
peoples.' But there is little agreement abou! how Romaniza!ion is
I Acculturation: I use the working definition ofR. L. Bee, Patterns and Processes:
(New York, 1974), 96: 'modifications within cultures resulting from contacts with
alien life ways'; see also H. G. Barnett et aI., 'Aeculturation', American Anthl'opo-
logist, 56 (1954), 973-1000; 1. H. F. Bloemers, 'Acculturation in the RhineJMeuse
Basin in the Roman Period: Demographie Considerations', in J. e. Barrett, A. P.
Fitzpatrick, and L. Macinnes (eds.), Barbarians and Romans in North-West Europe:
Fl'om the Later Republic to Late Antiquity (Oxford, 1989), 178 (a 'continuous
process of interaction between two or more autonomous culture systems and the
resulting change'). For current thinking about Romanization, see especially M.
Millett, The Romanization 0/ Britain: An Essay in Archaeological Interpretation
(Cambridge, 1990), with the review by P. W. M. Freeman, "'Romanisation" and
Roman Material Culture', JRA 6 (1993), 438-45. A number of important essays are
coJlected in Barrett, Fitzpatrick, and Macinnes (eds.), Bal'barians and Romans; T. F.
e. Blagg and M. Millett (eds.), The Early Roman Empire in fhe West (Oxford, 1990);
R. Brandt and J. Slofstra (eds.), Roman and Native in the Low Countries: Spheres of
Interaction (Oxford, 1983); see also A. T. Fear, Rome and Baetica: Urbanization in
Southern Spain c. jO Be-AD IjO (Oxford, 1996); M. L. Okun, The Early Roman
Frontier in the Upper Rhine Al'ea: Assimilation and Acculturation on a Roman
Frontier (Oxford, 1989); the section entitled 'Roman and Native' in V. A. Maxfield
and M. J. Dobson (eds.), Roman Frontier Studies 1989: Proceedings 0/ the XVth
International Congress of Roman Fl'ontier Studies (Exeter, 1991); G. Woolf, 'The
Unity and Diversity ofRomanisation', JRA 5 (1992), 349-52; 'Beyond Romans and
Natives', W. Arch. 28/3 (1997), 339-50. On the Romanization of north Africa, see
especially M. Bimabou, La Resistance africaine ci la romanisation (Paris, 1976); 'Les
Romains ont-ils conquis I'Afrique?', Ann. ESC 33 (1978), 83-8; T. R. S. Broughton,
The Romanization of Africa Proconsularis (BaItimore, 1929); T. Kotula, 'Les Afri~
cains et la domination de Rome', Annales litteraires de [' Universite de Besam;on, 186
(197 6),337-5 2; P. Leveau, 'La Situation coloniale de l'Afrique romaine', An11. ESC
33 (1978), 89-92; H.~G. Pflaum, 'La Romanisation de I' Afrique', Vestigia, 17 (1973),
55-7 2; Y. Thebert, 'Romanisation et deromanisation en Afrique: Histoire decolo~
nisee ou histoire inversee?', Ann. ESC 33 (I978), 64-82 .
 M easuring Romanization
to be recognized,2 or even about what it means. The process seems
3 to the Roman citizenship and, over time, to the inner circles of the
generally to be understood to describe the adoption or imitation of
Roman ways of thought, behaviour, construction, and manufac-
ture 4 Characterization is perhaps necessarily inexact. The inade-
quacy of the historical record impedes analysis of indigenous
practice or sentiment, especially outside the literate and mostly
urban, native 61ites who wanted to advertise their connections with
the governing power, and for whom Roman material culture repre-
sen ted a means of maintaining or of enhancing the prestige of their
own positions. 5
lt might even be questioned whether Romanization can properly
be said to be a function of cultural change. J. F. Gilliam onee
remarked that being a Roman (like being an American) was largely
a 'matter of law, not of culture' 6 He was correct, I think, at least at
one level-the requirements of the Roman citizenship were nar-
rowly and alm ost wholly juridical in nature. lt is likely, too, that
Roman cultural attributes were gradually modified, in ways that
are now probably unrecoverable, as more provincials were admitted
2 Cf. D. 1. Breeze, 'The Impact of the Roman Army on North Britain' in Barrett
Fitzpatrick, and Macinnes (eds.), Barbarians and Romans, 232: 'we ne~d to defin~
bettel' what we meall by "Romanisation" and how it might be expected to manifest
itself archaeologically'.
3 See also P. 0rsted, Roman Imperial Economy and Romanization (Copellhagen,
19 85),16; G. Woolf, 'Monumental Writing and the Expansion ofRoman Society in
the Early Empire', JRS 86 (1996), 37; 'Beyond Romans and Natives', 339.
4 Cf. 1. C. E.dmondson, 'Romanization and Urban Development in Lusitania', in
Blag~ ~nd MI.llett (eds.), Early Roman Empire, 153; 'a gradual change affecting
provillclal soclety, caused by the adoption (at least at an elite level) of the main
strands of Roman practices of government, law, language, dress and culture'; W. S.
Hanson, 'The Nature and Function of Roman Frontiers' , in Barrett, Fitzpatrick,
and Macinnes (eds.), Barbarians and Romans, 58: 'the acceptance and adoption of
Roman civilization and its values'; N. Mackie, 'Urban Munificence and the Growth
ofUrban Consciousness in Roman Spain', in Blagg and MilIett (eds.), Early Roman
Empire, 179; 'the assimilation ofthe Latin Ianguage, of Roman political, legal, social
and economic practices, of Roman architecture, artifacts and technological skilIs'.
5 Cf. Millett, Romanization 0/ Britain, 2J2: the Romanization of Britain's pro-
vincial eIites was 'largely indigenous in its motivations, with emulation of Roman
w~ys and styles being first a means of obtaining or retaining social dominance, then
bemg used to express and define it'; see also T. F. C. Blagg and M. MilIett
'Introduction', in Blagg and Millett (eds.), Early Roman Empire, 3; Fear, Rom~
and Ba~tica,. 28; A. P. Fitzpatrick, 'The Uses of Roman Imperialism by the Celtic
Barbanans 111 the Later Republic', in Barrett, Fitzpatrick, and Macinnes (eds.),
Barbarians and Romans, 31. Advertising connections: S. L. Dyson, The Creation 0/
the Roman Frontier (Princeton, 1985), 86.
6 'Romanization of the Greek East The Role of the Army', Bulletin 0/ the
American Society 0/ Papyrologists, 2 (1965), 66.
Measuring Romanization 77
governing elite. The end-product, it may be supposed, will have
been what David Mattingly has called a 'cultural cocktail', whose
'specific ingredients and proportions are now difficult to distin-
guish'.7 At the same time, however, it cannot be denied that there
was an identifiable Roman cultural matrix, at least through the first
and part of the second eenturies AD, defined by language and
custom, perhaps also by artifacts and architectural forms 8
It has been objected, too, that the very term 'Romanization' is
misleading, in so far as it implies a unilateral absorption of Roman
culture9 I should not want to deny that it may sometimes be
necessary to speak also of, say, 'Africanization' (perhaps especially
from the third century AD). W But it must be admitted that the cross-
cultural adoption of material and other cultural attributes in north
Africa was mostly one-sided, unequal both in degree and in kind.
What is far less certain is whether historians can solve the
methodological and inferential problems inherent in measuring
the extent and effect of Romanization. The main obstacle is the
virtual absence of evidence for the aspirations and intentions of
the acculturated; put another way, we cannot know (or expect ever
to know) the motives of the Romanized non-elites. n It is because
7 Mattingly: Tripolitania (Ann Arbor, I994), 38. Cr. P. van Dommelen, 'Colonial
Constructs: Colonialism and ArchaeoJogy in the Mediterranean' , W. Arch. 28/3
(1997), 309, 319-20 (on 'hybridization'); Fear, Rome and Baetica, 269; Woolf,
'Be~ond Romans and Natives', 341.
See also D. Braund, 'Introductioll: The Growth ofthe Roman Empire (241 BC-
AD 193)', in D. Braund (ed.), The Administration o/the Roman Empire (241 BC-AD
193) (Exeter, 1988), 10-II.
9 1. H. F. Bloemers, 'Acculturation in the Rhine/Meuse Basin in the Roman
Period: A Preliminary Survey', in Brandt and Slofstra (eds.), Roman and Native,
161. For acculturation as a reciprocal process, see especially H. 1. M. Claessen,
'Kinship, Chiefdom and Reciprocity; On the Use of Anthropological Concepts in
Archaeology', in Brandt and Slofstra (eds.), Roman and Native, 217; Millett, Roma-
nization 0/ Britain, 2; 'Romanization; Historical Issues and Archaeological Inter-
pretations', in Blagg and Millett (eds.), Early Roman Empire, 37. I do not agree wirh
1. C. Barrett ('Afterword: Render Unto Caesar', in Barrett, Fitzpatrick, and
Macinnes (eds.), Barbarians and Romans, 236) that we ought to abandon the study
of acculturation (even more generally of the relations between Roman alld native),
on the grounds that it requires 'evolutionary assumptions'.
JO Cf. Fear, Rome and Baetica, J03 ('Iberianization').
I I R. MacMullen, Changes in the Roman Empire: Essays in the Ordinary (Princeton,
1990),63-4; see also Millett, 'Romanization: Historical Issues', 38. Stilliess survives
to describe the purposes of those who, it may be assumed, rejected 01' resisted the
Romanization of their culture.
 Measuring Romanization
ofthewant of a written record that almost all ofthe models that have
been developed to assess Romanization (in general if not quantifi-
able ways), like Ramsay MacMullen's 'odd little list' of 'gods, pots
and Latin', are, in a broad sense, archaeological in nature, both in
the materials and in the methods they employ." Most of them, it
will be suggested, are, for one reason or another, inadequate.
My position, which is elaborated below, is that a model of
Romanization should, first, describe the aceulturation (or imita-
tion) of ideas or artifaets that are, in anthropologieal terms,
'embedded' in the intrusive (that is, Roman) culture, and therefore
not readily acculturated, or, alternatively, the substitution of
Roman ways or artifaets for concepts or objects that are
'embedded' in the indigenous culture; seeond, it should measure
the aeeulturation of those ideas and/or artifacts across a broad
social range of the indigenous eulture (that is, not just among the
literate elites); and, third, it should describe their aceulturation in
ways that are quantifiable aeross large sampIes. '3
SOURCES AND METHons
The evidence that survives to illustrate the nature of indigenous
cultures in the western provinces of the empire is, in the words of
Fergus Miliar, 'relatively slight, disparate and ambiguous'.'4 And
what little there is describes almost exclusively the habits of the
wealthy and mainly urban provincial elites, who are likely to have
absorbed Roman eultural attributes most readily, if for no other
reason than that they possessed a greater capacity to aequire the
products of Roman material eulture. '5 The culturallife and appe-
tites of ordinary provincials are virtually undocumented. '6
12 MacMullen: Changes, 61.
'3 On 'embedding', see S. E. van der Leeuw, 'Acculturation as Information
Processing', in Brandt and Slofstra (eds.), Roman and Native, JI-41, esp. at 32 .
J4 MilIar: 'Local Cultures in the Roman Empire: Libyan, Punic and Latin in
Roman Africa', JRS 58 (1968), 126.
15 Cf. Fear, Rome and Baetica, 275; MacMullen, Changes, 42, 46, 59 ('what
natives could not afford could never become apart of their lives').
16 Cf. 1. F. Drinkwater, 'For Better 01' Worse? Towards an Assessment of the
Economic and Social Consequences ofthe Roman Conquest ofGaul', in Blagg and
Millett (eds.), Early Roman Empire, 214, on Roman Gaul. It may be remarked, too,
that the little evidence we possess for the cultural life even of provincial elites
Measuring Romanization 79
It is impossible therefore to reconstruet now mueh of what Gr~~
Woolf has called 'the human experience' of Roman Impenahsm.
Gaps in the historical reeord are likely to co~eeal a eomplex
eultural reality, according to which the tYPlcal mdlgenous response
to eonquest and oecupation may be loeated at some llldefinable
point along a continuum bounded by the extremes of those who
eagerly embraced Roman values and those who rejected them. ,8 In
other words, it is diffieult to deseribe perfeetly those who were, III
varying degrees, 'partially' Romanized. It may even be suspeeted,
'th MacMullen that there existed a seeond world beneath the
W
one deseribed I, in our Roman (and Romamzlllg .. ) sourees, wholly
.
unclassieal in nature, perhaps an aetual majority of the populatIOn
of the provinces. '9 It is, it might be added, a world of whleh we are
afforded only an occasional glimpse, in north Afnean lllvocatIOns
of the dU Mauri, for example, perhaps also in the eunous httle
Arvernian statuettes of 'wet-nurses' .20 The ~rocess tha~ we call
Romanization is readily identified in the matenal r~eord, III depos-
its of Roman-made dinnerware beyond the Rhllle, III badly wntten
Latin inscriptions at the edges of the empire. But It IS far from
eertain that we possess the tools that might accurately measure It.
lt is in the absence of any real evidence, then-perhaps because
of it-that it seems now to have become fashionable to argue, or
really to assert, that ordinary provincials are likelyto have emu-
lated the recorded values and practiees of the ehtes; as Cohn
Haselgrove has put it, 'gradual emulation of elite behavIOur by
describes the views and sentiments only of male society. In other and somewhat
better documented societies (e.g. in the American West), cross-cultu~aI ~~n~cI~s
appear often to have been monopolized by indigenous. males: R. . ttc:,
'Introduction: Revisionism and Regionalism', in R. D. ~Itchell.(ed.), Appa!achwn
Frontiers: Settlement, Society and Development in fhe Premdustrtal Era (Lexmgton,
Mass., 1991), 9. Perhaps the same can be assumed of the Roman world.
17 Woolf 'Unity and Diversity', 350. , ,.
18 Cf F~ar, Rome and Baetica, 61; Y. Le Bohec, La Troisiem~ Left.lOn Auguste
(Pans,. .I 989) , 28 , 518',B. D
. Shaw, 'The Structure of Local Soclety m the Early
Maghrib: The Elders', The Maghrib Review, 16 (1991), 18.
19 MacMullen: Changes, 62. . 1
zo DU Mauri: see Benabou, La Resistance africaine, 3°9-30, esp. at 320-1, see ~ s~
G Camps 'Quisontles Dii Mauri?', Ant. afr. 26(1990), 131-53; E. w. B. Fen~resf? ~1l
Mauri and Dii Patrii' Latomus, 27 (1978), 5°7-16. Statuettes: H. Verte.t, Re 19~O?
opulaire et rapport ~u pouvoir d'apn!s les statue~tes d'argile sous l'~mplre rom~m :
in A. Daubigney (ed.), Archeologie ef rapports SOClaUX en C.au1e (Pans, 1984), ~8. 97,
see also 1. Webster, 'Necessary Comparisons: A Post-Colomal Approach to RehglOus
Syncretism in the Roman Provinces', W. Arch. 28/3 (1997), 33 2-4.
 80 Measuring Romanization
other sections .of the population, especially the upwardly mobile
elements, provlded the motor for the more widespread changes in
archllectural fashIOn and material culture which we conceive of as
Romanization':" The model, which has been given perhaps its
fullest expressIon by Martin Millett, may be summarized as
follows: because the Romans preferred not to administer the
provmces dl;ectly (mainly as a way ofreducing costs), they empow-
ered native ehtes to govern their own societies, provided that they
dld so broadly m accordance with Roman principles and with a
Roman-style constitution that provided, at the very least, for the
handmg over of tax revenues to the central government; the local
ehtes were happy to be co-opted and eager both to acquire and to
dIsplay the material symbols of Roman culture because their iden-
tification with Roman military and political authority served to
stre~gthen then already privileged position; 'progressive emula-
tIon of their acqnired values at lower levels of the social hierarchy
was 'self-generating'. 22
It is necessary to insist, however, that there is no real evidence
for 'self-generating' or any other sort of emulation of the beha-
VlOur or tastes of the provincial elites. It cannot be assumed that
ordmary provincials shared broadly in the aspirations of the lit-
erate . class. What is needed is a standard of measurement that
descnbes the acculturation of Roman valnes and artifacts across
a wlde ra,nge of the social structure of indigenous culture.
. MIllett s model may be said to be inadequate for other reasons.
FIrst, II cannot really be said to measure Romanization, because it
provldes no tool of quantitative analysis. MacMullen has remarked
that it is 'self-evident' that there was some acculturation in each of
the provinces of the empire: the real question is how much. '3
Second, hke many others, Millet!'s model aims at identifying the
21 Haselgrove: 'T.he Romanization of Belgic Gau!', in Blagg and Millett (eds.),
Early Ron;.an. /fn:P1re, 45; see also I. Ferris, 'Shopper's Paradise: Consumers in
Roman Bntal.n, In P. Rush (ed.), Theoretical Roman Archaeology: Second Confer-
~nce Proceedmgs ~Aldershot an~ Brookfield, Vt., 1995), 137. Cf. B. D. Shaw,
Autonomy and Tnbute: Mountam and Plain in Mauretania Tingitana', ROMM
41~2 (19 8?), 82 n. 3, wh<;, notes that the co-option oflocaI elites is often understood
to be the normal paradI.gl~ of pr~vjncial integration into the Roman system', and
sU~fest? th~t ,the mo~ei ~s sure!y lll. need of some iconoclastic revision'.
M~lIett. RomaOlzatlOn: Hlston.cal Issues', 38. See also B. Kurehin, 'Romans
and Bnton~ on th~ ~orthern FrontIer: A Theoretical Evaluation of the Archaeo-
logy of Reslstance , III Rush (ed.), Theoretical Roman Archaeology 12 5
23 MacMullen: Changes, 61. ' .
Measuring Romanization 81
assimilation only of Roman material culture: Roman techniques of
production and/or Roman-made or Roman-style goods. It seems
to me that the process of Romanization cannot be descnbed ade-
quately other than by distinguishing rigorously between the accul-
turation, on the one hand, of things, and, on the other, of the Ideas
that go with them. It has been said recently that Romanization is
'archaeologically demonstrable'.'4 But some at least of what we
might like to know about its effects is archaeologically untraceable:
post-conquest changes in patterns of land ownershlp, for example,
or in education and in the use of native languages. 25 And archae-
ology can tell us nothing at all about the reasans why some pro-
vincials adopted Roman ways. ,6
It seems often to be taken for gran ted that provincials will have
appropriated as many characteristics ofRoman cuIture as they could
afford. But adoption of a conqueror's way oflife is hardly a umversal
feature of pre-industrial empires. '7 In fact, it could be argued that
there was little about Roman culture that was inherently desirable,
perhaps only, in MacMullen's eslimation, 'hot baths, central heat-
ing, softer beds, and the ple~sures of wine'." It h~s been su~gested
recently that north Afncans may have found much that was attrac-
live' in Roman culture, for example, 'in technology and sepulchral
architecture'.2 9 What sort of superior technology did the Romans
have to offer to the peasants and pastoralists of the Maghreb?
Roman habits and ideas cannot be said to have been adopted
because they were self-evidently snperior to the indigenous. A
Roman-style name and clothes, at least a rudimentary knowledge
of Latin-these were among the prerequisites of advancement, m
army life or in the bureaucracy. It was Tacitus' understanding
(Agr. 21) that Britons who adopted Roman ways dld so because
they were eager to be promoted (hanoris aemulatio). It may be
questioned whether those who embraced Roman custom merely
because it was expedient to do so (that is, to enhance thelr own
position) can really be said to have been acculturated.
24 Barrett, 'Afterword', 235. 25 Breeze, 'Impact of the Roman Army', 227·
26 Cf. MacMullen, Changes, 63, 295 n. 3I.
27 Woolf, 'Unity and Diversity', 352. 28 MacMullen: Changes, 63·
29 D. B. Saddington, 'The Parameters ofRomanization', in Maxfield ~nd Dobs?n
(eds.), Roman Frontier Studies 1989, 416; cf. P. G~rnsey, 'Rome's Afn~a~ EI?-Plre
under the Principate', in P. Garnsey and C. R. Whlttaker (eds.), Impenahsm In the
Ancient World (Cambridge, 1978), 253·
 82 Measuring Romanization
It seems to me that what we ought to be trying to identify is the
acculturation of ideas or values that were rooted or, in anthropo-
logleal terms, 'embedded' in Roman culture. A Roman artifact that
carried with it some religious significance, for example, will have
been more difficult to acculturate than, say, a pot 30 Put another
way, the most acculturated among the indigenous will have been
those who deviated furthest from the norms of what anthropolo-
gists would call 'the most exclusive orders of structured activity' in
their own cuhure (so, conversely, the least aceulturated will have
been those who eonformed most closely to the same norms)3'
It is my position that many of the techniques which have been
employed to assess the aeculturation of indigenous societies in the
provinces are imperfect, because they da not measure, in ways that
are quantifiable aeross a wide range of provincial soeiety, either the
acculturation cf values Cf concepts that were rooted in Roman
culture, or their displaeement of beliefs and ideas that were
embedded in the indigenous eulture.
UNWORKABLE MODELS
The provineial adoption of Roman arehitectural forms is often
taken to be an index of aeculturation. So the Romanization of
Britain is said to be 'demonstrated more obviously and more
permanently by its arehiteeture than by any other aspect of its
2
material eulture' 3 We are told that 'the most charaeteristie
phenomenon' of Romanization is tbe provineial town that was
organized on astreet grid and equipped with 'substantial
Roman-style publie buildings', includingjora, basilicae, baths, tem-
pIes, amphitheatres, and theatres. 33 The Romanization of Britain is
3° van der Leeuw, 'Acculturation', 32.
3' See B. P. Dohrenwend and R. 1. Smith, 'Toward a Theory of Acculturation'
Southwestern
2
Journal 0/ Anthropo!ogy, 19 (1962), 35-6.
3 T. F. C. BJagg, 'Art and Architecture', in M. Todd (ed.), Research on Roman
Britain /960-89 (Londan, 1989), 210.
33 Millett, Romanization oi Britain, 69. Cf. Thebert, 'Romanisation et deromani-
sation', 69. on th~ bu.ilding of rural ~at~-complexes in north Africa. For fora, baths,
and theatres as mdlces of RomamzatlOn, see also P. MacKendrick, The 'North
A/rica? Stones Sp~ak (Chapel Bill, NC, 1980), 216-17; M. SpeideI, 'The Roman
Armym North Afnca', JRA 5 (1992), 404 (it was the 'highly romanized' provincials
not the 'country bumpkins', who built amphitheatres, capitoIia, circuses, and the~
atres); cf. H. Elton, Frontiers o/the Roman Empire (Bloomington, Ind., 199 6),41.
Measuring Romanization 83
said to be indicated also by the construetion of heated hornes and
of reetangular timber and clay buildings that functioned as both
house and ShOp.34 So, too, the building of Roman-style residences
in the eountryside is thought to be 'an important index of Roma-
nization', or at least of the Roman aspirations of their owners. 35
Aeeording to Keith Branigan, the construction of Roman-style
villas indicates 'a reasonable degree of Romanisation'; their own-
ers, he suggests, ean be understood tohave adopted 'to some
degree a Roman life-style'3 6 What, it might be asked, is a 'reason-
able' degree of Romanization?
The real question is whether Roman styles of arehiteeture were
adopted (or imitated) for any reason other than ease, eomfort, or
expediency. Put anotber way, can the adoption of Roman arehi-
teetural forms be assumed to indicate 'the aeeeptance of a set of
Roman values'?37 What provineial would not have preferred to live
in a heated horne? The urban elites who paid for the eonstruetion
of Roman-style publie buildings did so both as a way of advertising
their own importance and in the expectation that they would
thereby win the favour of the governing power. 38 In the case of
34 Heated homes: K. Branigan, 'Celtic Farm to Roman Villa', in D. Miles (ed.),
The Romano-British Countryside: Studies in Rural Settlement and Economy (Oxford,
1982), i. 95. Timber and day buildings: Millet~, Romanizatio~ 0/ Britain? 107. Cf.
Blagg 'Art and Architecture', 206: the five- or slx-roomed corndor house Illustrates
the 'g~owth of a more broadly based Romanized rural society'. ,
35 MilIett, Romanization 0/ Britain, 195; cf. 117-19. See also M. Fulford, The
Economy of Roman Britain', in Todd (ed.), Research on Roman Britain,. 187; N .. 1.
Higham, 'Roman and Native in England North of the Tees: Acculturatlon and 11s
Limitations', in Barrett, Fitzpatrick, and Macinnes (eds.), Barbaria~s ~nd Romans,
154; R. Hingley, 'Roman Britain: The Structure of Roman !mpenahsm .and, t~e
Consequences of Imperialism on the De~elo1?ment of a Penpheral, Provl~ce, m
Miles (ed.), The Romano-British Countryslde, 1. 32; A. H. A. Hogg, InvaSIon and
Response: The Problems in Wales', in B. C. Burnham and B. B. Johnson (eds.),
Invasion and Response: The Case 01 Roman Britain (Oxford, 1979), 285; S. D.
Trow, 'By the Northern Shores of Ocean: So:ne Observations .on Acculturation
Process at the Edge of the Roman World', In Blagg and MllIett (eds.), Early
' Roman Empire, 112-13. Romano-British villas are surveyed i~ I. Hodd~~ and
M MilIett 'Romano-British Villas and Towns: A Systematlc AnalysIs, W
Ar~h. 12 (1980), 69-76. 6
3 Branigan: :Cel.tic Farm', i. 81.
37 R. F. 1. Jones, 'A False Start? The Roman UrbamzatlOn of Western
Europe', WAreh. 19 (1987), 50: the building of Roman-style towns in Britain
shows 'the acceptance 01 a set 01 Roman va/ues, which ex.t~n.ded beyond the
construction of grand public buildings to paying for the actlVltles that went on
inside them, like games in the amphitheater' (my italics). See also Fear, Rome
and Baetica, 170. . . .
3 8 Cf. Fear, Rome und Baetica, 268 (on Roman-style pubhc works 10 Spam).
 Measuring Romanization
public buildings, it is sometimes impossible even to distinguish
between what is Roman and what is not. It has been suggested
that the pre-Roman public buildings at Leptis Magna, which have
not been preserved, were probably 'already very largely classical' in
39
style. At the very least, it must be admitted that the building of
large pnvate houses and villas teils us nothing about the aspira-
tlOns of those who were not of the landed elite40
Art, too, is an inadequate standard of measurement, and for
much the same reason. It cannot be disputed that the use of stone
as a medium of expression in Roman Britain is evidence of 'the
weight ofRoman fashion', or that the new figurative style that was
adopted at Maktar from the time of Trajan was broadly of Roman
. . . 4' ' .
lllsp,ratlOn. The problem, as Mdlett saw, IS that the material
record cannot distinguish between the tastes of artists and crafts-
men and those of their patronsY All that is really indicated is
exactly . what we might have expected to find, that weaIthy
provlllcml patrons wanted to express themselves publicly in
ways that were identifiably Roman.
Urbanization is also sometimes made to serve as a model of
Romanization. It seems to be widely agreed that 'towns, and the
monumentalization of towns, are a potentially revealing index to
the spread of Romanization in a particular area of the Roman
Empire'.43 It is a view that can be challenged on at least two
grounds. 44 First, if urbanization is understood to mean the con-
391. B. Ward-Perkins, 'Pre-Roman Elements in the Architecture of Roman
Tr~~olitania', Jn F. ,E G~dallah ,(e.d.), Liby~ i~ History CBen?h~zi, 1971),104.
Cf. ~ramgan, Celh.c Farm, 1. 95; the thmly-spread bUlldmg of a thousand or
tW? .fashlOnable, Romamsed, bungalows' does not indicate 'profound change in the
Bntlsh landscape' .
4) Stone; P. Salway, The Frontier People 01 Roman Britain (Cambridge, 19 5),27.
6
Maktar: A. M'Charek, Aspects de revolution demographique et sociale d Mactaris
auxZ lIe et IIIe siecles ap. J C (Tunis, 1982), 57- 8.
. 4 , MilJett: Romanization 01 Britain, 112-13, 1I 7. See also Freeman, 'Romanisa-
hon,441.
43 Edmondson, 'Romanization', 151; cf. 152; urbanization can be a 'useful index
to the extent to which the various parts ofthe province [Lusitania] were affected by
Romanization'. Less explicitly, Trow, 'Northern Shores', 103.
44 I can make even less sense of the theory that the Romans plan ted or encour-
aged settlements as an 'instrument' of Romanization: so Dyson, Creation, 116; see
also 0rsted, Roman Imperial Economy, 357 Ca 'means' of Romanization). Are we to
understand that the Romans b~i~t towns explicitly for the purpose of providing
exemplars of Rom~n ways of hvmg? Or does Romanization here (as often else-
where) serve as a kmd of euphemism for 'pacification'?
Measuring Romanization 85
struction of urban (or urbanized) centres, it cannot be taken to
be evidence of Romanization unless it can be shown-and this is
precisely what is often not demonstrable-that Roman-era towns
were constructed where there were no pre-existing (that is,
indigenous), proto-urban or nucleated settlements. The problem
may be expressed another way: how are we to distinguish in the
archaeological record (either of western Europe or of north
Africa) between town-building that was a direct consequence of
Roman conquest or contact, and that which may have been a
result of larger, that is, not specifically Roman, trends toward
urbanization?45 It has been said of the frontier-zone in Roman-
era Algeria, for example, that 'the city, and by extension, ~rban
civilisation' are not attested before the second century AD 4 But
there is, as C. R. Whittaker has pointed out, a 'strong likelihood'
that many of the small towns of Roman-era north Africa-va.r-
iously identified in the sources as castella, turres, pyrgOl-were III
fact either indigenous settlements or Romanized towns built on
indigenous sites. 47 Second, in so far as urbanization seems often
to be taken loosely to mean the development of Roman-style
towns it ought to be admitted that what is really being measured
is the' adoption of urban architectural forms, not urbanization.
It is no more likely that the acculturation of the provinces is
indicated by the spread of what is now often called 'the epi-
graphie habit'. It has been said of north Africa that the. 'habit
of inscription' which is attested in its smaller commumtles lS
evidence of the adoption of Roman 'urban cultural forms'.4 8 It
may be agreed that the practice of inscription (and perhaps
especially of commemoration on stone) was a chiefly Roman
45 The idea is borrowed from R. Reece, 'Romanization: A Point of View', in
Blagg and MilIett (eds.), Early Roman Empire, 31-3; see also ~aselgrove, 'R?n~a­
nization', 53. On the Iron Age origins of nudeated settlements m Roman Bntam,
see Jones, 'A False Start?', 47; Trow, 'Northern Shores', 110-11. For Lower
Germany, see 1. H. F. Bloemers, 'Lower Germany: Plura consilio quam vi.: .Pr?t?-
Urban Settlement Developments and the Integration of Native Commumtles , m
Blaiig and MilIett (eds.), Early Roman Empire, 72-86, esp. at 84·
4 E. W. B. Fentress, 'Forever Berber?', Opus, 2/J (1983), 162-3.
47 Whittaker: 'Land and Labour in North Africa', Klio, 60 (1978),352-3. See also
Leveau 'La Situation coloniale', 91; Mattingly, Tripolitania, 41; S. Raven, Rome in
Alrica, '3rd edn. (London and New York, 199.3), !OO-I. Mattingly: Tripolitani~, 160,
concludes that the foundation for the urbamzatlOn of north Afnca was PUDle, not
Roman. 8
4 Fentress, 'Forever Berber?', 163.
 86 M easuring Romanization
habit. But the survival of 'Libyan' inscriptions indicates that it
was not exclusively Roman.49
Another model of Romanization makes the adoption of Roman
or Roman-style names an index of acculturation. In that names are
culturally determined, it cannot be disputed that the substitution of
Roman für native farms is indicative of Romanization. 50 It is
undoubtedly true that Latin translations of north African names
(Donatus for Muthun, for example) demonstrate a 'Romanizing
and assimilationist spirit'5' The virtual disappearance of north
African nomenclature at Maktar in the second century AD is unde-
niable evidence ofits acculturationY And the adoption ofRoman-
style names in the provinces is one ofthe very few indices ofRoma-
nization that is recorded across large sampIes. But there is no way to
determine how many provincials took Roman-style names (in any of
the provinces), because it is impossible to distinguish systematically
between Roman (or Italian) immigrants and Romanized indigenes
on Latin inscriptions,53 Was the Iulius Saturninus who was comme-
morated at Lambaesis (eIL 8.3741), for example, an Italian immi-
grant, or a north African who had adopted Roman-style names, or
perhaps the Africanized son of a Roman immigrant?54 H.-G.
Pflaum once calculated that ofthe 1,269 free-born men and women
recorded on the inscriptions of Castellum Celtianum, fully 9 6.7 per
cent (1,227) had Roman or Romanized names. 55 But all that can
49 'Libyan' inscriptions are collected in 1. B. Chabot, Recueil des inscriptions liby-
ques (Paris, I940).
5° MacMullen, Changes, 60; see also D. 1. Buck, 'Prontier Processes in Roman
Tri'pol~tan~a', in D. 1., Buck and D. 1. Mattingly (eds.), Town and Country in Roman
Tr!foiltama: Papers m Ifonour ofOlwe~ Rackelt (Oxford, 1985), r86.
L. A. Thompson, Same ObservatlOns on Personal Nomenc1ature in Roman
Af~ica', Nigeria and the Classics, IO (1967-8), 57. On the name Donatus see 1.
KaJanto, The Latin Cognomina (Helsinki, 1965), 18. ' 59 Benabou, La Resistance africaine, 331-76; Edmondsoll, RomamzatlOn, 153,
5~ ~'Charek, Aspects de l'evolution, 59, 169: 'une romanisation poussee'. See also
Ma~tJ~gly, Tripolitania, 58: the substitution of Roman for Punic names in Tripoli-
tama IS attested from the last quarter of the Ist cent. AD.
53 MacMullen, Changes, 60; see also M. Brett and E. W B. Fentress, The Berbers
(05~ford: 199 6), 53~ Whittaker, 'Land an~ Labour', 343 n. 65.
D(IS) . M( antbus) . S( acrum) I IulIUS' Sa / turninus· Vi I xi! . An[n Jis. XL .
U / nus· Sextic: / Fusänila· Co(niunx). For Saturninus as a popular cognomen
am~mg Romamzed north Africans, see Benabou, La Resistance africaine, 502;
KaJanto, Latin Cognomina, 18, 55.
55 'Remarques sur l'onomastique de Castellum Celtianum', in E. Swoboda et al.
(ed.s.), Carnuntina: Ergebnisse der Forschung über die Grenzprovinzen des römischen
ReIches: Vorträge beim internationalen Kongress der Altertumsforscher Carnuntum
1955 (Graz and Cologne, I956), 134.
Measuring Romanization 87
really be said ofhis figures is that the un-Romanized proportion of
the population at Castellum Celtianum cannot be proven to have
risen above 3 per cent. 56
It has long been recognized that the adoption of Roman reli-
gious practices is evidence of the assimilation of ideas that w~re
culturally determined. 57 Romano-British cults, for example, wlth
'their dear fusion of Roman and Celtic', can be said to provide 'a
measure of just how Romanized Britain became' 58 Alternatively,
evidence of the continued worship of native deities may indicate
resistance to Romanization, or its weakness. 59
In north Africa, a number of indigenous gods were at least
partly Romanized. Shadrapa, for example, came to be identified
with Bacehus, Melkart with Hereules:o The loeal, mostly urban
elites seem generally to have been eager to embrace Roman reh-
gious ideas and practices 6 ' Tbe eult of the Capitoline triad, the
emperors, and the city of Rome is attested even at Mena'a, in the
heart of the Aures mountains." But there is good evidenee also of
an abiding attachment to indigenous beliefs. So Saturn, a rather
transparent disguise for Punic Baal, appears to have been enor-
mously papular, especially among soldiers and the (rural) poor63
Other indigenous (perhaps ethnie) gods like the dii Mauri, dU
Magifae, and dii Macni continued to be worshipped well into the
period of the Roman occupation, and, it may be suspected, long
after64
56 Cf. MacMullen, Changes, 60.
57 For the model, see Blagg and Millett, 'Introduction' , esp. at 3; Saddington,
'Parameters of Romanization', 414.
58 R. 1. A. Wilson, JRS 82 (1992), 291; cf. Bloemers, 'Acculturation: Preliminary
Survey', 170, on the Romanization of indigenous gods in Lo,;er Gert?an'y., .
MacMullell, Changes, 294 n. 24.
60 See also Ravell, Rome in Africa, 145.
61 P A Fevrier 'Religion et domination dans l' Afrique romaine', Dialogues
. . ,
d'histoire ancienne, 2 (1976), 310. 6~ See F entress, 'Forever B erb er.,.
?' 163
.'
63 On the popularity of Saturll, see especially M. Le Glay, Saturne afncam:
Histoire (Paris, 1966); see also Benabou, La Resistance ufricaine, 370-6. :or
soldiers, see Y. Le Bohec, Les Unites auxiliaires de ['armee romaine en Afnque
Proconsulaire et Numidie sous le Haut-Empire (Paris, 1989), 180.
64 For the dii Mauri, see above, n. 20. The dii Magifae are attested at Tinfadil
Hr. Metkides (CIL 8. 16749 = ILAlg I. 2977 = ILS 4493), the dii Macni at
Rusicade (eIL 8.8023 = 19981 = ILS 4I36); see also B. D. Shaw? 'Rural Markets
in North Africa and the Political Economy of the Roman EmpIre', Ant. afr. 17
(198!), 52 n. 6.
 88 Measuring Romanization
Some parts of these processes can be identified in the mate-
rial record, in the changing forms of temple structures and ilf
cult-images, for example, and in votive inscriptions, which occa-
sionally record the public beliefs of a section of indigenous society
that was below, though perhaps not much below, the urban elites
(nothing can tell us about the sentiments of the chronically
poor)65 But there are serious methodological obstacles to measur-
ing the Romanization of indigenous religion. Cult-images, for
example, are notoriously difficult to categorize systematically.
And the inscriptions cannot be used to trace changes in religious
practices or beliefs, because they are nowhere preserved in large
numbers over the whole of the Roman era.
Coinage is sometimes also taken to be a standard of provincial
66
acculturation. It cannot be denied that the use, say in south-
east Britain, of inscribed coinage of a type that is indisputably
'Romanizing' is indicative of at least some measure of 'material
Romanization'.67 However, finds of Roman coinage at provincial
sites cannot always be taken to be evidence of a change in local
practices or habits, because they may indicate nothing more than
the presence or proximity of Roman soldiers or traders. 68 It is far
from certain, too, that monetization itself indicates that provin-
cial societies-the Netherlands, for example-were 'more or less
romanized':9 because it is neither attested nor inherently likely
that not using coinage was an embedded characteristic of
provincial culture. And even where it can be shown that indigenous
societies began to use coinage in the Roman period, or dramatically
65 Temples and cult-images: A. C. King, 'The Emergence of Romano-Celtic
Religion', in Blagg and Millett (eds.), Early Roman Empire, 222. On the under-
representation of the poor on inscriptions, see especially 1. C. Mann, 'Epigraphic
C~~sciousness', JRS 75 (1985), 204.
Für the model, see Fear, Rame and Baetica, 57; 0rsted, Roman Imperial
Economy, 15-
67 C. HaseIgrove, 'The Later Iran Age in Southern Britain and Beyond', in Todd
(ed~, Research on Roman Britain, 13.
6 Cf. M. L. Okun, 'Pluralism in Germania Superior', in Maxfield and Dobson
(eds.), Roman Frontier Studies 1989, 437; C. R. Whittaker, Frontiers 01 the Roman
E11Jfire: A Social and ~conomic Study (BaItimore, 1994), 123.
R. Brandt, 'A Bnef Encounter along the Northern Frontier' , in Brandt and
Slofstra (eds.), Roman and Native, 137. Cr. I L. Davies, 'Soldiers, Peasants and
Markets in Wales and the Marches' , in T. F. C. Blagg and A. C. King (eds.), Military
and Civilian in Roman Britain: Cuttural Relationships in a Frontier Province (Oxford
1984), 109, II2.
Measuring Romanization 89
increased their use of it, it can rarely be demonstrated that the
coinage itself is unquestionably Roman (or Romanizing)7 0
Much the same can be said of the appearance of Roman-style
graves and grave-monuments in the provinces7 ' MacMullen has
suggested that the Romanization of north Africa is indicated. by
what he considers to be a change in indigenous mortuary-practIce,
according to which the dead were buried 'no longer under mere
heaps of stones, doubtfully identifiable, but under shaped monu-
ments that gradually came to be inscribed'7' However the stele-
type gravestones that were the predominant form of burial-
monument in north Africa through much of the second century
AD are not really so very different from the pre-Roman, undeco-
rated, funerary steles-what MacMullen calls 'heaps of stones'-
that have been found at a number of sites in eastern Algena (most
especially in the Aures and Hodna mountains, and in the region
around Sitifis)73
Perhaps the most widely employed method of assessing Roman-
ization attempts to track the provincial distribution of goods that
were of Roman manufacture or style 74 An example, which is
widely paralleled: the incorporation of flat plates and footrings
in locally produced wares in Upper Germany in the period
7° See N. Purcell, 'The Creation of Provineial Landscape: The Roman Impact on
Cisalpine Gaul', in Blagg and Millett (eds.), Early Roman Empire,.20-I. ,
1
7 Cf. MaeMullen, Changes, SI; D. I Mattingly and R. B. Hlt.ehner, Roman
Afriea: An Arehaeologieal Review', JRS 8S (199S), 186 (on erematIon).
2
7 Changes, 294 n. 29. Cf. M'Charek" 1-spects .de l'evolution, 193 (t?e ~n­
Romanized at Maktar eontinued to prefer simples plerres tombales sans decor ).
73 Benabou, La Resistance alricaine, 494-S; see also P. A. Fevrier and R. Guery,
'Les Rites funeraires de la necropole orientale de Setif', An!. air. IS (1980), 91-124;
R. F. 1. Jones, 'Burial Customs of Rome and the Provinces', in I S. Wacher, (ed.),
The Roman World (London and New York, 1987), ii. 829. Cf. Haselgrove, Lat~r
Iron Age', 17, on south-east Britain in the late Iron Age and Ist cent. AD; w.her~ 11
appears to have been only the loeal eIites who adopt~d what he calls a mlllon~y
Romanising burial rite'. The typology of north Afnca.n fune,ra.l monum~.nts IS
described in I-M. Lassere, 'Recherehes sur la chronologie des epitaphes palennes
de l'Africa', Ant. afr. 7 (1973), 120-2; Le Bohec, La Troisieme Legion Auguste, 85;
Les Unites auxiliaires, 14.
74 The model is neatly summarized in MacMullen, Changes, 292 n. 9; cf. M.
Fulford 'Roman and Barbarian: The Economy of Roman Frontier Systems', in
Barrett,' Fitzpatrick, and Maeinnes (eds.), Barbarians and Rorr:ans, 8S. For Rom.an
goods as 'symbols' of Romanization, see R. Rebuffat, 'Au-dela des camps romams
d'Afrique mineure: Renseignement, contröle, penetration', ANRW 21I01z (1982),
S02, S09; see also Dyson, Creation, 220 (on pottery); Hingley, 'Roman Britain', i. 21;
'
MacMullen, Changes, 42.
 90 Measuring Romanization
between 100 BC and AD 69 indicates that loeal potters had
adopted Roman styles. 75
An important advantage of the model is that the goods found at
provincial sites can be counted, and, in some cases, like that of
pottery, dated (at least in a general way). The problem is that there
is usually no way of determining whether the produets were
adopted or imitated beeause they were thought to be culturally
superior or prestigious (in so far as they were symbolic ofthe ruling
power), or simply because they were cheaper, better made, or better
adapted to the requirements of large-seale produetion. 76 The pre-
senee of Roman-style button-and-Ioop fasteners and glass bangles
at native sites in nortllern Britain may be evidence, not of cultural
assimilation, but of 'fast sales talk'. 77 Finds of Roman pottery and
brooehes in the so-ealled 'buffer zone' that lay beyond the Rhine
may indieate nothing more than 'the presenee or proximity of the
Romans'7' And because the objects may have been taken up with-
out the eoncepts that aceompanied them, we must, as Whittaker
has reeently put it, 'be careful not to exaggerate or to assurne that
Roman artifaets turned transborder folk into Romans'79
A more sophistieated version of the same model maintains that
the adoption or imitation of certain kinds of Roman artifacts
75 M. L. Okun, 'An Example of the Process of Acculturation in the Early Roman
Frontier', Oxford Journal o[ Archaeology, 81I (19 89),46--7_
76 MacMul!en, Changes, 42; so, tao, Freeman, 'Romanisation', 444. A modern
analogy: OWllll1g, say, a Japanese-made car does not make one Japanese.
77 Breeze, 'Impact of the Roman Army', 230.
78 L. Hedeager, 'Empire, Frontier and the Barbarian Hinterland: Rome and
Northern Europe from AD 1-400', in M. Rowlands, M. Larsen, and K. Kristiansen
(eds.), Centre and Periphery in the Ancient Wor/d (Cambridge, 1987), 127. On the
'buffer zone', see also Hedeager, 'A Quantitative Analysis of Roman Imports in
Europe North of the Limes (0-400 A.D.), and the Question of Roman~Germanic
Exchange', in K. Kristiansen and C. Paluden~Müller (eds.), New Directions in
Scandinavian Archaeology (Studies in Scandinavian Prehistory and Early History,
I; Copenhagen, 1978), 191-216, esp. at 204-7.
79 Concepts: van der Leeuw, 'Acculturation', 29. Whittaker: Frontiers,
12 5. R.
DeneIl, , '!he Hunter~Gatherer/Agricultural Frontier in Prehistoric Temperate
Europe, m ~. W. Green and S. M. Perlman (eds.), The Archaeology 0/ Frontiers
and Boundanes (Orlando, Fla., 1985), 113-40, argues that trade is not in itself
su~fic!ent to p.roduce social. change. See also Higham, 'Roman and Native', Ij3-
4: m mterpretmg the matenal remains of Roman provincial culture 'much can be
explained without recourse to models of culture change'; C. M. WeHs, 'The Pro~
blems of Desert Frantiers', in Maxfield and Dobson (eds.), Roman Frontier Studies
1989,4 80: Roman goods were 'superimposed, but without necessarily changing very
much at the profaunder level of existence'.
M easuring Romanization 9I
indicates change in indigenous customs that were eulturally spe-
cific, for example, in eating or drinking habits80 So the presence of
sigillata in Upper Germany is said to be evidence of the reglOn's
acculturation, because it implies the use of individual sets of
dishes, where it had been, according to Poseidonius (Athenaeus
4. 15 I), La Tene custom to share drinking vessels and other items
of tableware. " Similarly, the adoption of Gaulish, 'and thus essen-
tially Roman', dietary preferences in south-east Eritain is thought
to indieate the 'material Romanisation' of tbe region's elites." It is
likely enough that the provincial elites of the empire wanted to
emulate Roman fashion in the preparation and consumptIon of
food and drink 83 What cannot be demonstrated is that the impor-
tation of Roman products or their imitation had any effect on the
habits of the mass of ordinary provincials, that is, in MacMullen's
wards, 'of poorer people,.8.
The adoption of Latin, especially as a spoken language, or of
Roman styles of dress, can also be said to describe the assimilation
of attributes that were culturally determined. 85 Eut neither process
is capable of being measured across large sampIes. A~d neither can
be shown to have operated outside the provlllcial ehtes (hke the
wealthy Carthaginians who donned togas or fixed their hair in the
styles favoured by the women of the imperial family)86 So though
80 Okun, 'Example', 47, ja; see also P. Middleton, 'The Roman Army and ~on~­
Distance Trade' in P. Garnsey and C. R. Whittaker (eds.), Trade and Famme In
Classical Antiqufty (Cambridge, 1983), 7j: 'the m~terjal ass.emblage of :'importe.d"
fine ware pottery and wine amphorae' represents a potential body of ldea~ whl.ch
flowed along the trade rautes'; M. Struck, 'Analysis of Social and Cultural Dlverslty
on Rural Burial Sites in North~Eastern Raetia', in Rush (ed.), Theoretical Roman
Archaeology, 74.
8r Okun, 'Pluralism', 437. The indigenous population ap~ears als? to have
adopted the use of Roman~style small bowls and cups (Okun, Example, 49)·
82 Haselgrove, 'Later Iran Age', 13. . .
83 See also Trow, 'Northern Shares', 103, describing how post~co~lques~ Br~tam
imported fine tablewares, mortaria, and ceramic and metal vessels (mc1udlllg Jugs,
cups, and strainers). 84 MacMull~n: ,Cha.n~es, ~3; cf. 62.
Ss Latin: see Blagg and Millett, 'Introduction', 3; M. Hemg, RehglOn III Rom~n
Britain', in Todd (ed.), Research on Roman Britain: 21.9; ?rsted, Roman Impenal
Economy, Ij; Saddington, 'Parameters of RomamzatlOn, 414; cf. Edmondso.n,
'Romanization', Ij2, on the continued use of native langua~es. Dress: J.. EadJe,
'Peripheral Vision in Roman History', in J. D'Arms and J. Eadle (eds.), Ancl,ent and
Modern Essays in Hanor 0/ Gerald F. Else (Ann Arbor, 1977), 232; Okun, Plural~
ism', 437 ('a strang indicator of cultu~e'). .. ., .
86 Raven, Rame in A/rica, 146. Pumc dress lS dlS.cuss~d III M. Charek, A~pec!s ~e
!'evolution, 38-9. J. P. Wild, 'Roman and Native m Textde Technology, m
 92 Measuring Romanization
the Punic script seems to have disappeared from north African cities
by the end of the second century AD, a bastardized form of the
language ('neo-Punic') continued to be spoken in the countryside.87
The Romanization of provincial society is said to be indicated
also by the adoption of Roman-style municipal government
. c~nnection between the status of towns and the extent to which
(magIstrates an d town-councIls),. 88
and by the development of
the fiaminate, which Paul MacKendrick calls 'a potent instrument
of Romamza . t'IOn , .89 I t seems to me to be unlikely that many
provmclals even of moderate means ever served as decurions Of
as priests of the imperial cult. For much the same reason I cannot
agree with Peter 0rsted that the enfranchisement of p~ovincials
was a 'means' of acculturating the provinces. 90 The main purpose
(and effect) of the granting of the Roman citizenship was the co-
option of local elites. Few ordinary provincials were given the
citizenship, and then normally only as areward for conspicuous
service to Rome, until after about AD 150, when its value was
eroded by the wholesale enfranchisement of towns and even of
entire regions. In fact, the principal benefits that accompanied the
citizenship-the right of appeal and the right in Roman law to
receive bequests from other Roman citizens-are unlikely to have
been of much importance to most provincials in any period.
MacMullen is surely right in thinking that the so-called Constitutio
Antoniniana of AD 212, which gave the citizenship to virtually all of
the empire's free-born inhabitants, is likely to have met with 'the
absolute indifference of the masses'.9'
Another method that has been used to assess Romanization
seems to have nothing at all to do with acculturation. It has
Burnham and Johnson (eds.), Invasion and Response, 127, concludes that Celtic
styles of dress changed very little in Roman Britain.
:; See Raven,. R~me in Afri~a, 146. 88 Fentress, 'Forever Berber?', r 63.
~~cKe~dnc~. Nort,h ,Alr/can Stones, 51; see also X. Dupuis, 'La Participation
des veterans a Ia Vle mumclpale en Numidie meridionale aux He et IIIe siecles' in C
~epe~ley (ed.), Actes du IVe colloque international sur l'histoire el l'archeolo~ie d~
1Alrtque d~ lY,0rd (Strasb~urg: 5-9 avril, 1988) (Paris, 1991), 34J-54; T. Kotula,
Culte provmcml ~t ro~all1SatlOn: Le Cas des deux Mauretanies', Eos, 63 (1975),
3~9-4.07. M. Chnstol, Rome et les tribus indigenes en Mauretanie Tingitane',
LAjI~lca roma~a.' 5 (1988), 316, maintains that the flaminate played an important
role 1ll Romamzmg the Iocal elite at Gigthis.
9° 0rst~d: Roma~ Imperial Economy, 357. A similar argument is advanced in
M. Dondm-Payre, Recherches sur un aspect de la romanisation de 1'Afrique du
Nord: L'Expansion de la citoyenneU~ romainejusqu'a Hadrien', Ant. all'. 17 (19 81 ),
1
93-132. 9 MacMullen: Changes, 60.
Measuring Romanizatian 93
sometimes been supposed, perhaps especially of north Africa, that
the Romanization at least of its urban centres is indicated by their
promotion to the status of, say, municipia or calaniae,9 2 But there
is as Robert Broughton recognized long ago, no demonstrable
they were Romanized (or, it might be added, their size, wealth, or
location)93
It has been suggested also that the recruitment of provincials
into the Roman army functioned as a 'highly effective and rapid
acculturation mechanism',94 According to Yann Le Bohec, the
auxiliaries who served in north Africa adopted 'Ie genre de vie,
les institutions et le droit de Rome'95 It might be questioned
whether military service really made natives into Romans 96
What is reasonably certain is that the number of provincials
recruited into the Roman army was never more than a tiny
proportion of the indigenous population.
Tables 3.1 and 3.2 describe the geographical origins of the
soldiers of the Third Augustan legion whose birthplace (origo) is
recorded. They indicate that as late as the time of Hadrian, less
than one-fifth of the soldiers were from north Africa. And with the
exception of those born 'in the camps' (castris), most of the Afri-
can-born recruits were from the large urban centres of Africa
Proconsularis (especially Carthage) or the first-century AD army-
towns like Ammaedara and Theueste 97 Excluding those born
castris, not a single soldier appears to have been recruited from
the frontier-zone over the whole of the period to AD 235 98 Table
9" So, at length, Pflaum, 'La Romanisation de l' Afrique'.
93 Broughton: Romanization 01 AjI-ica, 128. Size, etc.: B. D. Shaw, 'Archaeology
and Knowledge: The History of the African Provinces of the Roman Empire',
Florilegium, 2 (1980), 36-7.
94 Bloemers, 'Acculturation: Preliminary Survey', 203.
95 Les Unites auxiliaires, 189.
6
9 See Gilliam, 'Romanization', 66. He concludes (67) that auxiliary soldiers were
'Roman in some degree' (my italies). See also Fear, Rome and Baetica, 266; Mat-
tingly, Tripolitania, 168, on the 'Punicized' or 'Libyanized' Tripolitanian garrison of
the 3rd cent. AD.
97 See also B. D. Shaw, 'Soldiers and Society: The Army in Numidia', Opus, 2/1
(1983), 144. The distribution of legionary origines at the end of t~e 2nd cen~. AD is
mapped in Le Bohec, La Troisieme Legion Auguste, 519. On cas~rzs, see espectally F.
Vittinghoff, 'Die rechtliche Stellung der canabae legionis und dIe Herkunftsangabe
castris', Chiron, 1 (1971), 299-318. .
8
9 Pace Brett and Fentress, Berbers, 55, who maintain that the soldters of the
frontier-zone 'came from successful peasant families in the neighbourhood'.
 94 Measuring Romanization Measuring Romanization 95
Table 3· 1. The origins 0/ recruits 10 Ihe Third Augustan legion
3.1 indicates also that the proportion oflegionary recruits who are
DateIperiod Percentage of recruits said to have come from the camps (castris) increased significantly
over the course of the second century AD, from 4 per cent in the
Africa, Castris: African All Africans Non- time of Hadrian, for example, to 34 per cent in the time of Anto-
excluding army camps Africans ninus Pius. Their absolute. number, however, is likely to have been
cas/ris very small-Iess than 100 recruits will have been needed in any
given year to replace soldiers discharged from the legion. 99 And
Hadrian 15 4 19 81 because most of those who were born in the camps were probably
Antoninus Pius 32 34 66 34 legionaries' sons, they are likely, as Gilliam has put it, already to
AD 225 64 36 100 0 have been acculturated 'to some extent'. 100
Far less survives to describe recruitment to the auxiliary units
~ourc:e:. Y Le Bohec, La Troisieme Legion Auguste (Paris, 1989), 70. The main
ll1SCnptIOos are eIL 8. 18084-5; AE (19 06), 124. stationed in north Africa. Mainly because Tacfarinas is known to
have served in the auxilia, it has been suggested that they were
recruited heavily from the Musulamii. IOI There is, however, no
good evidence to show that even a majority of the north African
auxiliaries were of African origin. Of the twenty-six auxiliary
Table 3.2 The origins 0/ African-born recruits 10 (he Third Augustan legion soldiers whose geographical origin is recorded, just three were
from north Africa. 102
Site Nurnber of recruits Much of the surviving evidence then points in the same direc-
tion: military service is unlikely to have had any significant effect
AD I4-II5 AD 116-235 Total AD 14-235 % AD 14-235 on indigenous culture. The 20,000-25,000 soldiers stationed in
Ammaedara
north Africa (legionaries and auxiliaries) represented perhaps
0 23 23 2.8
Carthage 0.3-0-4 per cent of the region's population (6-8 million?). <03 Simi-
12 108 120 14·8 larly, the roughly 150,000 auxiliaries who were in active service in
Cillium 0 13 13 1.6
Cirta 3 45 48 5·9 99 See Shaw, 'Soldiers and Society', 138-40.
Hadrurnetum 26 27 3·3 100 Gilliam: 'Romanization', 70.
Lambaesis 0 322 322 39·8 101 Benabou, La Resistance africaine, 124; M. Rachet, Rome et les Berberes: Un
Sicca Veneria 0 13 13 probleme militaire d'Auguste cl DiocIetien (Brussels, 1970), 161; see also J.-M.
1.6
Thamugadi 0 31 Lassere, 'Le Recrutement romain et les Musulames', in Lepelley (ed.), Actes du
31 3. 8
Thelepte 0 12 IVe colloque international, 299-31 I.
12 1.5 102 Le Bohec, Les Unites auxiliaires, 173. Cf. J.-M. Lassere, 'Remarques onomas-
Theueste 7 33 40 4·9 tiques sur la liste militaire de Vezereos', in W. S. Hanson and L. J. F. Keppie (eds.),
Thysdrus 0 I I I I Roman Frontier Studies 1979 (Oxford, 1980),955-75, whose analysis of the names
Utica 1.4
3 9 12 1.5 of the soldiers reeorded as serving at Vezereos on the Tripolitanian fron tier in the
Other 6 131 137
period between AD 199 and 2II (R. Cagnat, A. Merlin, and L. Chatelain (eds.),
16·9 Inscriptions latines d'Afrique (Tripolitaine, Tunisie, Maroc) (Paris, 1923), no. 27)
Total 32 777 809 suggests that most of them were of African origin; see also D. J. Mattingly, 'Libyans
and the Limes: Culture and Society in Roman Tripolitania', Ant. afr. 23 (1987), 82.
;ourc~s: G. Forni, Ill'eclut~~~ento ~e~le legioni da Augusto a Diocleziano (Milan, 6_8 million?: Raven, Rome in Africa, 88-9 (6-7 million); R. P. Dunean-
103

953)., y. Le Bohec, La TrOl:mme LegIOn Auguste (Paris 1989) 5'9-20' B D Sh Jones, 'Wealth and Munifieenee in Roman Africa', Papers of the British School
'Sold d S·
. ~ers ~n . o~lety:
The Army in Numidia', ' ,
Opus, 21I (1983), 144-6. The
,.. aw
'
al Rome, 31 (1963), 170 (8 million). The indigenous population of Aigeria at the
time of the Freneh eonquest (1830s) was about 3 million: P. M. E. Lorein,
pnnclpal mscnptlOns are eIL 8. 2564-9, 18068, 1808 4-7. Imperial Identities: Stereotyping, Prejudice and Race in Colonial Algeria (London
and New York, 1995), 2.
 Measuring Romanization
the provinces at any given time in the period befare Septimius
Severus were only about 0.3 per cent of the estimated population
of the provmces (44-54 million). '04 I suppose that there may have
been a sort of 'multiplier effeet' according to which auxiliary
soldiers' families will have absorbed at least some of their
Romanizing tendencies. But the empire-wide effect is likely to
have been negligible.
It might be supposed that the army itself played a role in
:trans~itting the Roman way of life to the provinces', through
bmldmg programmes', for example, or veteran colonies . .05 It
has been said recently that, exeept in Upper Moesia (where it
was an 'alien and intrusive element among the native population'),
the army served as a 'bridge' between the indigenous population
and the 'administrators, artisans, merchants and farmers' who
'followed [it] to the frontier' . .06 Roy Davies concluded that the
ordinary Roman soldier was, 'more than anyone else', responsible
for 'the development of romanisation'. lO7
It is likely enough, I suppose, that the army disseminated some
Roman ideas and practices in some parts of the empire (though
probably not as a matter of policy). It has already been remarked,
however, that the north African frontier-army appears to have
bmlt alm ost entirely far itself. And there is no evidence of any
kmd to show that It funetioned as a 'bridge' between the intrusive
and indigenous populations. MacMullen has said of the western
provinces that the amphorae and moulded red dishes that the army
manufactured 'brought the Roman way of life into the natives'
very hornes' . .08 But there is nothing to indicate that the north
African ~nny was engaged in large-scale production of pottery
for the clVlhan market. Like the army in Spain, Germany, and
Bntam, It probably obtained much of the kitchen-ware it needed
44-:-54 mi1Ji~n: .the figure iso deriv~d from K. Hopkins, Conquerors and Slaves:
4
10.
SoclOl0fJ. 1cal Studles in Roman HIstory, 1 (Cambridge, 1978), 1,68, who estimates the
p°Po~latI.on of the ~mpire as a wh?le at 50-60 ~i11ion, ~nd t~at of Italy at 6 million.
Mlddleton, Roman Army, 75. The eVldence IS reVlewed in Y Le Bohec
L'1::nee romaine sous le Haut~Empire (Paris, 1989), 531-72. ' or Peacetime Army', in W. Eck and H. Wolff (eds.), Heer und Int~gratlOnspohflk:
S. K. Drummond and L. H. Nelson, The Western Frontiers 0/ Imperial Rome
(Armonk
10
and London, 1994), 67-8.
7 Service in Ihe Roman Army, ed. D. Breeze and V A. Maxfield (New York
1989), 68. 108 Changes, 58:
Measuring Romanization 97
from loeal potters . .09 It can even be conjectured that Roman
military rule, by usurping some of the traditional functions of
native elites, may sometimes have eroded Iheir prestIge, and
thereby impeded their acculturation.' .0
It might be supposed also that legionary and auxiliary veterans
will have transmitted Roman ways and habits to the regions where
they settled.'" So the veteran colonies of north Africa are saidto
have been 'isolated cases of Roman influenee among the natIve
communities'. II2 The veterans were undoubtedly more accultu-
rated than soldiers still in service. But they are likely also to
have been few in number.
Through much of the first and second centuries AD, there were
usually twenty-five legions in the provinces, each of whleh eon-
sisted probably of about 5,000 men, who were normally reqmred to
serve twenty years. If the mortality rate in service was, as Peter
Brunt has suggested, approximately 40 per cent, the number of
legionaries discharged annually will have been somewhere 111 the
order of 3,750. "3 If we assume also, with one recent study, that the
uniform annual death rate among veterans aged 40-60 was 5 per
cent and that the average age at enlistment was 20, we may
esti~ate that at any given time there were roughly 45,700 retired
. 114
legionaries aeross the whole of th e empIre. .. .
Aceording to Tacitus (Ann. 4. 5), the number of auxIlmry soldlers
in the provinces was roughly equal to the number of leglOuanes. If
we assume a standard term of service of twenty-five years and a 40
per cent rate of mortality in service, we can estimate that the number
10
9 See K. Greene, 'Invasion and Response: Pottery and the Roman Army', in
Burnham and Johnson (eds.), Invasion and Response, 99·
!IO Cf. MilleU, 'Romanization: Historical Issues', 39.
!II SO Elton, Frontiers, 57.
112 C. R. Whittaker, 'Rural Labour in Three Roman Provinces', in P. Garnsey
(ed.), Non~Slave Labour in the Greco-Ron:an. World (Cambridge,. 198?), 75. See also
W. S. Hanson, 'Administration, UrbamsatIOn and AcculturatIOn m the Roman
West', in Braund (ed.), Administration 0/ the Roman Empire, 53-4·
113 Brunt: Italian Manpower 225 B.C.-A.D. 14 (Oxford, 1971), 339; see also Sh.aw,
'Soldiers and Society', 155 n. 28. 3,750: see B. Dobson, 'The Roman Arm~: Wart.n~e
Die römischen Militärdiplome als historische Quelle (Cologne and "ylenna, 1986), 14
n. 16. The figure assurnes that each legion d~scharged 150 soldlers annually; cf.
Shaw, 'Soldiers and Society', 138-40 (100 soldlers annually). .
!I4 Recent study: Drummond and Nelson, Western Frontiers, 188-9; the~ estl~
mate that 3.400 legionaries retired each year, and that there were 37.400 leglOnary
veterans in all of the western provinces.
 98 M easuring Romanization
of auxiliary veterans discharged annually was about 3,000. 05 If 5
per cent of the veterans aged 45-65 died each year, there will have
been, at any given time, roughly 36,600 auxiliary veterans. o6 The
total number of veterans in the empire then, at any given point in the
penod between Augustus and Septimius Severus, is unlikely to have
exceeded 82,300, or about 0.2 per cent ofthe whole ofthe provincial
populatIOn (44-54 million).
About 630 legionary and auxiliary soldiers are likely to have
been dlscharged annually in north Africa (assuming a 40 per cent
mortahty rate,~;nong the 25,000 or so soldiers), perhaps 270 in the
fronller-zone. If 5 per cent of them d!ed each year, there will
have been, at any given time, roughly 7,700 veterans across the
whole of north Africa.' '8 Expressed another way, there were about
seven veterans for every 100 square miles. I 19 The veterans were
not, of course, distributed evenly across the region. In fact, so
many of them appear to have taken up residence in the civilian
s.ettlements that grew up around army bases, or in veteran colonies
hke Thamugadi and Diana Veteranorum, that there are likely to
have been many parts of north Africa in which there were in some
periods, no veterans at all. 120 '
It seems to be widely agreed that veterans usually settled in the
areas where they had been stationed. '" It is hard to believe, how-
ever, that many north African veterans will have chosen to settle in
the semi-arid lands of the frontier-zone. 122 Outside Lambaesis,
II5 See ~Iso Dobson, 'R?man Army', I4 n. 16 (his figure for legionaries with 25
years' servIce may be apphed to auxiliaries).
. lT6 Cf. Drummond and Nelson, Western Frontiers, 188 (13,000 auxiliary veterans
m the western provinees).
'" Cf• G ar~sey, ' R
118 ?mes' Af' ,
nean~mpire,346n.3I.
Cf. Whlttaker, Rural Labour, 75 (he eoncludes that there were probably
2,000 north African veterans in each generation).
119 The 'flat-map' area of Roman north Afriea was about I ro,ooo square miles;
see above, Ch. 2 n. I2I.
, 120 Cf. ~. ~a,hbou?i: 'Le~ Elites municipales de la Numidie; deux groupes:
etrr~~gers a la clte et veterans, ANRW 21IO/2 (1982), 673-81.
Se~ e.g. E. W ~. Fentress, Numidia and the Roman Army: Sodal, Militaryand
Econo'!llc Aspects oj the Frontier Zone (Oxford, 1979), 125; 1. C. Mann, Legion01Y
Recruttrnent and ':eteran S~tt~ement. during the Principate (London, 1983). There is
absolutely no basIs for behevmg, wlth Dyson, Creation, 273, that 'in most fron tier
area~, the largest group of rural settlers was probably the retired 01' discharged
soIdlers.'
122 See also Sh.aw, 'Soldiers and Soeiety', I40, who eoncludes that the evidence for
veteran-farmers m the area around Lamasba is 'most meager'.
M easuring Romanization 99
horne to weil over half of the region's attested veterans, very few
are recorded: nine, for example, at Lamigigga; seven each at
Theueste and Zarai; six (perhaps seven) at Casae; five at Diana
Veteranorum; four at Lamasba and at Calceus Herculis; two at
Mascula and perhaps also at Aquae Caesaris; just one at Ad
Maiores, and at Lambiridi. "3 Some inscriptions that record veter-
ans are no doubt undiscovered, especially in rural areas, and at
sites like Diana Veteranorum, where there has been little or no
excavation. And there will have been at least some veterans who,
for one reason or another, were buried without an inscribed
epitaph. But veterans seem also to be over-represented in the
epigraphic record: they are attested on almost 7 per cent (182) of
the 2,624 complete or mostly complete epitaphs that survive from
the fron tier-zone of Roman-era Algeria.
INTERMARRIAGE
Still another model of Romanization makes the marriage of
Roman and native an agent of"acculturation. 124 Intermarriage is
understood to have functioned as a kind of bridge between the
occupying and indigenous cultures, I25 at least in so far as any
children that resulted will have been culturally 'half-Roman'. <26
The model carries with it two important advantages. First, it
describes, in the case of north Africa, for example, both the
Romanization of indigenous culture and the 'Africanization' of
Roman provincial society. Second, the epitaphs and other inscrip-
tions that record marriages (and other less formal unions) are
12
3 Lambaesis: Fentress, 'Forever Berber?', 168 (127/234). The other figures are
taken from Shaw, 'Soldiers and Soeiety', 155 n. 31. A complete list of the veterans
recorded in the frontier-zone may be found in 1.-M. Lassere, Ubique populus:
Peuplement et mouvements de population dans l'Afrique romaine de la .chute de
Carthage ci la fin de Ia dynastie des Severes 046 av. C.-235 p. c.) (Pans, 1977),
174-89. The distribution of veterans in the region is mapped in Fentress, Numidia,
139·
12
4 See 1. A. Ilevbare, 'Family and Wornen in North Africa from the Fifth to the
First Century B.e.', Nigeria and the Classics, 10 (1967-8), 34; M'Charek, Aspects de
!'evolution, 107.
12
5 See e.g. Okun, 'Example', 51: acculturation on the Rhine frontier may have
occurred 'through the rnarriage ofRoman soldiers and civilians (traders, merehants,
etc.) to loeal women'.
126 'Half-Roman': Dyson, Creation, 273; see also GiIIiam, 'Romanization', 68.
 100 Measuring Romanization
preserved in large numbers. What has never been demonstrated
IS that mtermarnage occurred often enough in any part of the
empIre to have affected, in any significant degree, the acculturation
of the mdlgenous (or of the Roman) population. "7
The real obstacle to measuring the frequency of intermarriage is
methodologlcal. It was remarked earlier that on inscriptions it is
often ImpossIble to distinguish between, on the one hand, Roman
(or Itahan) Immlgrants and, on the other, provincials who adopted
Roman-style names. One solution is to divide those attested
I~stead mto two broad groups, which can, in most cases, be readily
dlstmgUlshed: on the one hand, those who are demonstrably un-
Romamzed; on the other, those who may be categorized either as
Roman cltlzens or as Romanized indigenes.
This scheme is used below (Chapter 4) to estimate how often
Roman(lzed) married un-Romanized in the frontier-zone of
Roman-era Algen~. What I am proposing to measure then, over
the several generatIOns represented in the surviving north African
mscnptIons, IS, not how often Roman married indigene (which is
hlstoncally unrecoverable), but the incidence of intermarriage
across cultural IdentItJes. My position is that the marriage of un-
Romamzed and. Roman(ized) defines acculturation almost as
closely as the UnIon of, say, north African and Italian.
7 Cf. -!'1attingly a.nd Hitchner, 'Roman Africa', 173: 'more work on intermar~
12
• und das die Soldaten des Prinzipats treffende Eheverbot' , in W. Eck and H. WoIfT
nage ... IS an essential prerequisite to the study of acculturation processes' .
4
Husbands and Wives zn
the Frontier-Zone
From at least the time of Claudius to that of Septimius Severus,
soldiers were forbidden to marry or, it seems, to be married
during their term of service.' The prohibition evidently did not
apply to officers, however. 2 And it did not prevent soldiers from
forming stable and, in at least some cases, long-lasting unions
with women whom they treated as their wives. The evidence is
mostly epigraphic~epitaphs, for example, on which a soldier
1 The ban is widely attested, expressly in BGU 114. I; see also P. Cattaoui 3-4
(text in L. Mitteis and U. Wilcken, Grundzüge und Chrestomathie der Papyruskunde
(Leipzig, 1912), ii. 372); O. Behrends, 'Die Rechtsregelungen der Militärdiplome
(eds.), Heer und Integrationspo!itik: Die r6mischen Militärdiplome als historische
Quelle (Cologne and Vienna, 1986), rr6-66. P. Garnsey, 'Septimius Severus and
the Marriage of Soldiers', California Studies in Classical Anfiquity, 3 (1970), 46,
argues that existing marriages were not dissolved by enlistment. But it is hard to
imagine that there could have been two classes of soldiers, one with wives, the other
denied the right to marry; see also S. Treggiari, Roman Marriage: fusti Coniuges
/rom the Time oi Cicero to the Time of Ulpian (Oxford, 1991),44. For Claudius, see
Cassius Dio 60. 24. 3 (AD 44). The introduction of the prohibition may go back to
Augustus, perhaps to 13 BC, when he appears to have made a number of changes in
the terms and conditions of military service: B. Campbell, 'The Man'iage of Sol~
diers under the Empire', JRS 68 (1978), 154; 1. E. G. Whitehorne, 'Ovid, A.A. I.
101-32, and Soldiers' Marriages', Liverpool Classical Monthly, 4 (1979),157-8; cf.
E. Birley, 'Before Diplomas, and the Claudian Reform', in Eck and Wolff (eds.),
Heer und fntegrationspolitik, 249. Its purpose is nowhere made explicit. But it may
be inferred from a letter that Hadrian sent to Rammius, prefect of Egypt, in AD 119
(BGU 140) that the authorities considered marriage to be inimical to military
discipline. For Septimius Severus, see Campbell, 'Marriage of Soldiers', esp. at
153; The Emperor and the Roman Army, 31 BC-AD 235 (Oxford, 1984), 302; R.
MacMullen, Soldier and Civilian in the Later Roman Empire (Cambridge, Mass.,
1963), 126; G. R. Watson, The Roman Soldier (London, 1969), I37; cf. Garnsey,
'Septimius Severus', 45.
2 Pliny, Ep. 6. 31. 4-6; Tacitus, H;st. I. 48, 4.5; cf. Epictetus, Ench. 3· 22. 79·
 !02 Husbands and Wives in the Frontier-Zone
Husbands and Wives in the Frontier-Zone I03
commemorates 01' i8 commem t db
as his 'wife'. 3 ara e y a woman who is described be10w, wou1d seern to indieate that the opposite is more 1ike1y to
It seems now to be widely b r d have been true-few of the soldiers or Roman(ized) eivilians who
stationed along the fron tier e ;eve that the soldiers who were are attested appear to have married indigenous women. About
indigenous women. G R v,(s 0 ten marned or eohabited with Lambaesis, for examp1e, it may be agreed with Brent Shaw that
the 'Ioeal women with 'h atsohn, for example, has written that 'all discernible patterns point to a strang element of marriage
w om t e soldlers formed between soldiers and other soldiers' daughters,8 The (se1f-
permanent associations were usuall of ' . more Of less
status at least in the f t' y peregnne [I.e. non-Roman] irnposed?) segregation of the Roman(ized) population at Lambae-
, fon ler afeas where mo t f h .
stationed' 4 Loeal s 0 t e soldlers were sis is even more evident when its marriage-patterns are compared
. wornen we are tald w '
canabae, and marriages b~tw h ' ere attracted to the to those recorded on the surviving, mostly eivilian, tornbstones of
even if they were contrar to een sU(~ warnen and the soldiers, Thubursieu Numidarum (mod. Khemissa), weil to the north ofthe
network of interrelations~ps bretgUlatlOns, so on began to weave a fron tier-zone, where the intermarriage of Roman(ized) and un-
the loeal population'. 5 Of the SO~d~::n the men of the fortress and Romanized seems to have been much more common.
has been said reeently that 'b th s who served III north Afriea it
usually contracted a eom IY e tJme they retrred they had
mOll- aw marn .h SOURCES AND METHODS
and settled down with her and th' age Wlt 6 a loeal woman,
stationed at Lambaesis we ar el~ e~I!dren'. The legionaries
the region'. 7 , e assure , cohabited with wornen of Almost nothing survives in the Iiterary sources to deseribe the
There is, however no evidence f k' intermarriage of Roman and north African. A single instanee
(or the Roman eivili~ns who took ~ any llld, to show that soldiers wou1d seem to be indicated, in De Bello Africo 19, where the
rautinely married indi p resldenee III the frantier-zones) anti-Caesarian offieer T. Labienus is said to have eonseripted an
genous women (lll north Af . . indeterrninate number of north Afriean 'ha1f-breeds' (hybridae) in
the other provinees) T h ' nea or III any of
ing epitaphs of the' AI e n:arr~age-patterns attested on the surviv- 46 Be: they were probab1y the offspring of Italian rnerchants who
genan rantler-zone, which are exarnined had married or eohabited with north Afriean wornen in the period
3 See also M. E. Snape 'Roman and Naf . Y .
before the eivil war9
V. A. Maxfield and M. 1. DObson (eds) Ive. ICI o~ the North British Frontier', in Not mueh ean be aseertained either from the so-ealled dip10rnas
th~ XVth International Congress 0/ R'o~~~m;n F~~~tler S.tudies J9 89: Proceedings of that were issued, probably from the time of C1audius, to veterans
Roman Soldier, 135; cf. Cam bell' ~?nflU Studle~ (Exeter, 1991),470.
probably formed liaisons with wo~ell ~f ~arn~ge of Soldlers', 154 ('most soldiers of the auxilia, navy, Praetorian Guard, and equites singulares. TO
veterans in M. Roxan 'Th D' t'b . p regnne status'). Much the same is said of
h' h ' e IS n utlOn of Roman MTt
p ISC.: e Studien, 12 (19 81 ), 270; A W van B . ' 11 ~ry Diplomas', Epigra- 8 Shaw: 'Soldiers and Society: The Army in Numidia', Opus, 2/1 (1983), 148 (his
Soldlers and Veterans', in M. Re~at:d (ed
1962), 1565.
)Ul'iI n, Some F~mllies Formed by Roman
., ommages a Albert Grenier (Brussels
italics). Much the same seems to be indicated by the experience of other historical
frontiers: e.g. cavalrymen in the American West in the late 19th cent. rarely married
5 S. K. Drummond and L H N I ' native women (P. Y. Stallard, Glittering Misery: Dependents of the Indian Fighting
(A[~O~k and Londo.n, 1994), 130. e son, The Western Frontiers 01 Imperial Rome Army (San Rafael, Calif., 1978), 69)·
. aven, Rome In Alric.·a 3rd edll (L d 9 Discussion in 1. A. Ilevbare, 'Family and Warnen in North Africa from the Fifth
!'1allton, Roman North Alrica' (Londo~ 1 on on aI?-d ~e.w York, 1993), 75. E. L. to the First Century RC.', Nigeria and the Classics, 10 (1967-8), 41-2; 'Some
,entered into cornmoll-Iaw liaisons with 'th~~8), ~8~ IS ,sImIlar: .m~ny of the soldiers Aspects of Sodal Change in North Africa in Punic and Roman Times', Museum
Romans alld Britons on the North F o~a gIrls. On Bntam, see B. Kurehin Alricum, 2 (1973), 31; see also L A. Thompson, 'Roman and Native in the Tripo-
Arch aeo Iogy 0 f Resistance' in P R ern h drontter'A.Theoret'lcaI E va nationI '
of the litanian Cities in the Early Empire', in F. F. GadaJIah (ed.), Libya in History
ond Conjerence Proceedings' (Ald~rs~~t ;~(/B Theorellcal Roman Archaeology: Sec- (Benghazi, 1971), 239.
The Romanization 01 Britain: An Essa in Ar rookfie~d, Vt., 1995), 126; M. Millett, 10 They were apparently not given to legionaries; cf. Camp bell, Emperor and the
1990),66 (many soldiers 'probably b Y .chaeologzcal InterpretaNon (Cambridge Roman Army, 442. On fleet diplomas, see especially G. Forni, 'I diplomi militari dei
7 T. R S Broughton Th R ecame mvolved with native wornen') , c1assiari delle flotte pretorie (inc1usi quelli dei c1assiari"legionari)', in Eck and Wolff
19 29), 137-8. , e omanization 01 Alrica Proconsularis CBaltimore, (eds.), Heer und Integrationspolitik, 293-321. The earliest extant diploma (eIL 16,
uo. I) was awarded in AD 52 to a man named Spartacus, who had served in the fleet
 1°4 Husbands and Wives in the Frontier-Zone
Initially, it seems, areward for meritorious service to Rome, they
came eventually (perhaps from the time of Titus) to be awarded to
those who had completed the requisite term of service, twenty-five
years, for example, m the case of auxiliaries," The distribution of
their find-spots suggests that they may have been issued normally
only to thoseveterans who intended to se!tle along the frontiers, "
where they mlght be expected to take up with women who were not
Roman citizens, '3 It may be, too, that they were given only to those
who agreed to pay for them, '4
The diplomas awarded the Roman citizenship to the veterans
and, until tbeir terms were altered about AD 140, to tbeir cbildren
and descendants (liberis posterisque). Tbe men were also given
conubium-tbe 'right of intermarriage'-with the 'wives' tbey bad
at the tIme the citizenship was granted to them, or with the first
women tbey married after they were discharged (conubium cum
uxo~ibus, quas tune habuissent, cum est civitas iis data, aut, siqui
caelzbes essent, cum iis, quas postea duxissent singuli singulas).I5
stationed at MisenuJ?; the, earliest s'!fviving auxiliary diploma (eIL 16, DO. 2) dates
to AD 54; see also Blrley, Befate Dlplomas', 249.
!, See 1. C. Mann, 'The Development of Auxiliary and Fleet Diplomas' Epigra-
phische Studien, 9 (1972), 234. ' record tbe names oftbe veterans' 'wives': CIL 16, no. 161, awarded
12 .Most of ~he auxiliar~ diplom~s have been found in the Danubian provinces
(I?acJa, ~oesIa, Pannoma, Raeha); see H.-J. Kellner, 'Die Möglichkeit von
~ucksc~l~ssen aus der Fundstatistik', in Eck and Wolff (eds.), Heer und Integra-
tlOnspolttlk, 246.
/3 Cf. V. A. Maxfield, 'Systems of Reward in Relation to Military Diplomas', in
E.ck and Wolf~ (eds.), Hee~ und Integrationspolitik, 43. This might explain also why
diplamas contlllued t~ b~ lssued (~o sailors, guardsmen, and the equites singulares)
e~en after the COnstltutlO Antonmiana of AD 212 had made Roman citizens of
":lrt~aI~y all of the empire's inhabitants; see also M. Roxan, 'Women on the Fron-
tIers,.lll ~axfield. a~d Dobson (e~s.), Roman Frontier Studies 1989, 466 . On the
Constl~u~1O 1ntonmwna, see especmlly F. Millar, 'The Date of the Constitutio
Antommana, Journal oj Egyptian Archaeology, 48 (1962), 124-31" C. Sasse Die
Constitutio Antoniniana (Wiesbaden, 1958); A. N. Sherwin-White, :The Tabula of
Banas.a a~d the ~~nstitutio Antoniniana', JRS 63 (1973), 86--98; H. Wolff, Die
Const!,utw Antommana und Papyrus Gissensis 40 I (Cologne, 1976), with the review
by MIliar, JRS 67 (1977), 235· The Iatest surviving auxiliary diplama dates to AD
2°3; see WEck and H. Wolff, 'Ein Auxiliardiplom aus dem Jahre 203 n. Chr.', in
EC1~ and Wolff (,eds.), He~r und Integrationspolitik, 55 6-75.
~. ~oxan, ??ServatlOns on the Reasons for the Changes in Formula in Diplo-
m~~ CIrca AD 140. ' III Eck and Wolff (eds.), Heer und Integrationspolitik, 265-6.
. !he quotatlOns are from eIL 16, no. 55, a diploma issued in AD I I 7 to an
auxlh~ry named. ~ogeti.ssa. It is not .entirely c1ear why the veterans' 'wives' were not
also glven the cltJzenshlp; see H. LIeb, 'Die constitutiones für die stadtrömischen
Truppen', in Ec~ and ~olff (eds.), Heer und Integraaonspolitik, 32 5; 1. C. Mann, 'A
Note on Conubmm', IbId. 18 7-9.
Husbands and Wives in the Frontier-Zone 1°5
Consequently, veterans could contract lawful marriages with
women who were not Roman citizens; even more remarkably, eXlst-
ing unions with non-Roman women were transformed into fuHy
val1'd marnages.
. ,6
For some reason, tbe wording of auxiliary diplomas was later
altered (apparently from November/Decernber, AD 140), in such a
way that veterans' existing children were no longer awarded the
citizenship." It cannot be tbe case that, as K. Kraft bad It, the
purpose was to discourage soldiers from formmg umons. wlth
women who were not Roman citizens, for tbe dlplomas contmued
to give veterans the rigbt to contract valid rnarriages with non-
Roman wornen. ,8 It rnight be supposed instead that tbe new tenns
were intended to promote auxiliary recruitment-perhaps a~xlh­
aries' sons were henceforth given tbe citizenship when they enhsted
(like legionaries' sons). '9
For reasons that are not at all clear, the number of diplomas
recovered from nortb Africa is disappointingly smalI: just one, far
example, from the wbole ofRoman~era Alger!: (another thirty-one
bave been found in Mauretama Trngrtana). And Just SIX of the
thirty-two north African diplomas (all from Mauretania Tingitana)
in AD 109 to a man named Bargati(s?), tbe husband ofIulia Deisata;
CIL 16, no. 169/73, given in AD 122 to M. Antomus MaxImus, wbo
had married Valeria Messia; eIL 16, no. I71 (AD 124), Issued to an
auxiliary whose name is now illegible; his wife's name is given as
( ... )a Sat( ... ); Roxan, Roman Military Dip/omas 1954-1977, no.
I I (AD 100/7), awarded to Hiern( ... ), wbo bad marrled Iaphna;
Roxan Roman Military Dip/omas 1954-1977, no. 18 (AD 114120),
on whicb the veteran's name is illegible; bis wife is identified as
16 See Gaius, Inst. I. 57.
. '
17 For the date, see Roxan, 'Observations', 271, 274. (The earhest extant dlploma
with the new wording-M. Roxan, Roman Military Diplomas /954-1977 (London,
1978), no. 39-dates to 13 Dec., AD 140.) A little more than half ?f t~e ~ata.ble
auxiliary diplomas (108/213) belong to the period AD 100--50. Thelr dlstnbutlOn
over time is plotted in Roxan, 'Observations', 284. .
/8 Kraft: Zur Rekrutierung der Alen und Kohorten an Rhem und Donau (Berne,
1951), 117-2 1. .. , .
19 See 1. C. Mann, 'The Frontiers of the Pnnclpate, ANRW 21I (1974), 516..... 17,
Roxan 'Observations', 278. I cannot explain why sailors' sons continued to be gIven
the citizenship; cf. Roxan, 'Observations', 275 (,stran,ge').. '.
20 Cf. Y. Le Bohec, Les Unites auxiliaires de l'armee romatne en Ajrzque P,ocon-
sulaire et Numidie sous le Haut-Empire (Paris, 1989), 8.
 106 Husb'ands and Wives in the Frontier-Zone
(... )a Rufina; and Roxan, Roman Military Diplomas 1978- 19 84,
no. 84 (AD 109), awarded to Sitalis, the husband of Iunia. It is
impossible to determine from so small a sampIe whether the soldiers
stationed in north Africa normally married indigenous women.
What the diplomas can be said to demonstrate is that the authorities
who issued them evidently expected that at least some auxiliary
veterans would marry wornen who were not Roman citizens.2r
A great deal more information is supplied by the many hundreds
of Latin epitaphs from north Africa which record the names of
husbands and wives. Most of those which have been recovered
from the Algerian frontier-zone (on which, see below) are published
in volume 8 ofthe Corpus Inscriptionum Latinarum (1881- ). Some
others which were unearthed subsequent to the publication of CIL
8, and which are included among the fron tier-zone epitaphs exam-
ined below, are published in volume 1 of Inscriptions latines de
l'Algerie (1922), which collects inscriptions from the province of
Africa P~oconsularis, including Theueste and its environs; in L.
Leschi, Etudes d'epigraphie, d'archeologie et d'histoire africaines
(1957); and in P. Morizot, 'Inscriptions inedites de l'Aures', Zeit-
schrift für Papyrologie und Epigraphik, 22 (1976). It is unfortunate
that volumes 3 and 4 of Inscriptions latines de l'Algerie, which were
planned as far back as the 1950S, have yet to be published; they
were meant to cover, respectively, the part of Roman-era Algeria
that was governed by the legionary legate ('Numidie militaire'),
and the province of Mauretania Caesariensis.22
Tbe epigraphic record of Roman north Africa is characterized
by an unusually high rate of inscription-survival, in part because
urban settlements were densely concentrated in the Roman era,
partly also because post-Roman settlement was generally less
extensive than in many other parts of the empire. '3 The bulk of
21 Cf. M. L. Okun, 'Pluralism in Germania Superior', in Maxfield and Dobson
(eds.), Roman Frontier Studies 1989, 435; P. Salway, The Frontfer People 0/ Roman
Britain (Cambridge, 1965), 30-I.
22 See L. Leschi, Etudes d'epigraphie, d'archeologie el d'histoire africaines (Paris,
1957), ~I-~. Val.. 2" whic~ ,:"as pu.biished in two parts, collects inscriptions from 'la
confedera.tlOn Cirteenne (mc1udmg Rusicade and Cirta), Cuicul (mod. Djemila)
an.d 'la tnbu des Suburbures'; none of the sites is in the frontier~zone. Vol. 3 was
sald to be ready for publication io 1989 (P. A. Fevrier, Approches du Maghreb
romain (Aix~en~Provence, 1989-90), i. 18).
23 R. P. Duncar:-~Jones, The Economy 0/ the Roman Empire: Quantitative Studies,
2nd edn. (Cambndge, 1982), 63, 361; he estimates that the survival-rate of 000-
funerary public monuments is in the order of 5%.
Husbands and Wives in the Frontier-Zone [07
the dated inscriptions, including epitaphs, appear to belong to the
period AD 100-25°.24 It may be assumed that the distribution over
time of the surviving epitaphs is representative of all epitaphs that
were erected, because no kind of accident either ofpreservation orof
excavation is likely to be specific to any given period. 25 Put another
way, there is no reason why any one period should be more under- or
over-represented than any other in the surviving record. .
Elizabeth Meyer has recently charted the distribution over tIme
of the epitaphs recovered at several nortb African sites. 26 Figure
4· 1 illustrates her methods by plotting the so-called 'epigraphic
curve' of the 672 dated epitaphs that have been recovered from the
Algerian fron tier-zone (they are listed in Appendix 4). Following a
technique developed by Ramsay MacMullen, and adopted by
Meyer, I have averaged over twenty-five-year periods the many
epitaphs that cannot be dated to any specific twenty-five-year
period (those, for example, which have been dated to the penod
AD 100-300).'7
Meyer's explanation of the epigraphic curves that she constructs
hinges on her understanding of the relationship between commem-
oration and the Roman citizenship. She appears to beheve that
only Roman citizens entitled to make a will valid in Roman law
were normally commemorated on Roman-style deceased-com-
memorator epitaphs. 28 And so an apparent increase in the number
24 Duocan-Jones, Economy, 65; cf. L. A. Thompson, 'Some Observations on
Personal Nomenc1ature in Roman Africa', Nigeria and the Classics, 10 (1967-8), 5 I.
25 On accidents ofpreservation, see Fevrier, Approches du Maghreb ror:zam, 1. 7?-
7; N. Mackie, 'Urban Munificence and the Growth of Urban ConsclOusne~s I.n
Roman Spain', in T. F. C. Blagg and M. Millett (eds.), The Early Roman Emprre In
the West (Oxford, 1990), 182. . .
26 'Explaining the Epigraphic Habit in the Roman En;pIre: T~e EVIdence of
Epitaphs', IRS 80 (1990), 74-96; on which, see D. Cherry, Re-figunng the Roman
Epigraphic Habit', The Ancient History Bulletin, 9 (I995), 14~-50.
27 MacMullen: 'The Epigraphic Habit in the Roman ~mpIre', AlP 103 (I982!,
241; adopted in Meyer, 'Epigraphic Habit', 82 n. 43. EpItaphs not dated by thelr
texts have been assigned the dates given to them in I-M. Lassere, 'Recherches sur la
chronologie des epitaphes palennes de l'Africa', Ant. afr. 7 (1973), 7-152, and Y. Le
Bohec La Troisieme Legion Auguste (Paris, 1989); see Appendix 4.
28 'Epigraphic Habit', 79: 'a Roman-type tombstone, in making manifest a
Roman legal relationship [i.e. heirship] even if couched on~y in the language of
the cornmernorator's affection, can also serve to make rna111fest the fact that the
deceased has acquired the right to create that relationship and imp.ose it~ obliga-
tions-that the testator: in short, possesses the fight to make a WIll valid under
Roman law'; cf. 83- So, too, 1. Morris, Death-Ritual and Social Structure in Classical
Antiquity (Cambridge, 1992), 168: the Roman citizenship was a 'necessary
 108 Husbands and Wives in the Frontier-Zone Husbands and Wives in the Frontier-Zone 109
200
Epitaphs dated to 25- Antoniniana (AD 212), which she believes devalued the Roman
year periods citizenship by granting it to most of the empire's inhabitants,
180 and in so doing made its pos session DO lünger worth announcing
• Average number of on tombstones. 3I
160 epitaphs The argument fails for a n!lmber of reasons. First, it is necessary
to insist that there was nothing to prevent non-Roman citizens
140 from being commemorated on Roman-style epitaphs. And there is
no good reason to believe that they were uniformly or even routi-
"
-" 120 nely excluded-how, in any event, could this have been accom-
il" plished? I believe, too, that the increase Meyer plots in the number
·ä
~ " of epitaphs at Theueste and other north African sites around AD
0 100
200 is in largo measure iIlusory, a product of the methods used to
""
.D
a0 date them.
80
Z Figure 4.2 employs Meyer's techniques to chart the epitaphs of
Theueste dated by Jean-Marie Lassere." Series I of Fig. 4.2 charts
60 the distribution over time of the seventy-eight epitaphs which
Lassere could not date to any specific twenty-five-year period,
40 and which I have averaged over twenty-five-year periods. Most
(fifty-nine; 76 per cent) of them can be dated no more dosely than
20 to the period AD 100-300. Series 2 of Fig. 4.2, which is reproduced
as Fig. 4.3, distributes the thirty-six epitaphs that Lassere assigned
0 to twenty-five-year periods. I have no reason to believe that any of
25 75 100 125 150 175 200 225 250 275 300 Lassere's dates are incorrect or that the technique of averaging
AD 1-300
widely dated epitaphs over twenty-five-year periods is in itself
unsound. What I am certain is untenable, because it produces
Fig 4. I. Dated epitaphs o[ the !rontier-zone resulls that are misleading, is the practice of charting an epi-
graphie curve that lumps together closely dated epitaphs with
of epitaphs in north Africa in the second century AD is said to be a those that can be assigned only to broadly defined periods. The
product of the extensIOn of the citizenship. '9 A sharp rise that she effect is to conceal two important truths (iIlustrated in Fig. 4·3):
charts In the number of epitaphs at Theueste around AD 200, for first, that the number of epitaphs that can be dated to any given
example (cf. FIg. 4.2), IS made a consequence of its promotion to twenty-five-year period is very smalI, and second, that the appar-
the status .of colony (colonia)3 D Similarly, what seems to be a ent rise in the number of epitaphs erected at Theueste both at the
reductlOn In the number of epitaphs erected in north Africa in beginning and at the end of tbe second century AD, and so also
the thlrd century AD is attributed to the issuing of the Constitutio what seems to be a reduction in their numbers in the third century,
is entirely a product of the methods according to which Lassere
c~ndri~n for erecting these tombstones'. Meyer believes also (85 n. 53) that almost
a 0 t e. ~ommemorators recorded on the surviving north African epitaphs were
Roman clhzens.
1
29 Meyer, 'Epigraphic Habit' 79 82 3 'Epigraphic Habit', 89, 95.
2
Lassere: 'Recherehes sur la chronologie'. I have excluded epitaphs that are said
or'Th~ue;te's promotion, see below, pp.
3
3° Ibid. 83-5. On the date 124-5_ to have been [auod 'in the vicinity' of Theueste.
 IIO Husbands and Wives in the Frontier-Zone Husbands and Wives in the Frontier-Zone 11 I

30
~ Series 2: epitaphs dated to 25- 25
year periods

N= 114
25 • Series 1: average number of N=36
epitaphs
20

20
"
"'II'" .:'l 15
'fr
~
15
·i
~

,0 ~

,0
""E" p"a
Z Z
0 10
10

5
5

o o -1-----,--,------,-----,-
25 50 75 IOD 125 150 175 200 225 250 275 300 25 50 75 100 125 150 175 200 225 250 275 300
AD 1-300 AD 1-300

Fig. 4.2 The epigraphic curve at Theueste Fig. 4.3 Epitaphs ai Theueste dated to 25-year periods

assigned all ofthe closely dated epitaphs to one oftwo twenty-five-
year periods, AD 100-25 or 175- 2 00.
. Mainly on the grounds that Dis Manibus Sacrum (a formulaic
InVOcatIOn of the gods of the underworld customarily abbreviated
D . M . S) began to appear routinelyon epitaphs at Theueste
sometime early in the second century AD ('debut du He siecle'), he
systemallcally assigned stele-type monuments ('dalles') on which
the formula is recorded to the period AD 100-25. And because he
believed that steles were replaced at Theueste by cippus-style
monuments ('autels') near the end of the second century AD ('fin
du He siede'), and that they in turn were replaced by burial-vaults
(cupulae; 'caissons') in the first half of the third century, he routinely
dated cippus-style epitaphs to the period 175-200.33 It may be said
therefore that charts like the one that is Fig. 4.2 are actually
plotting Lassere's methods. And because most of the dates he
produced are conjectural, the effect of charling them in the aggre-
gate is to pile one supposition upon another.
It is extremely doubtful, then, that a rise or fall in charts like
Fig. 4.2 can be linked to any specific historical event. It is entirely
possible that a change in policy (like the Co~stitutio Antoninian~)
or in a community's status (like the promotIOn of Theueste) WIll
33 Lassere 'Recherehes sur la chronologie', 120-2. Cf. Le Bohec, La TroisiCme
Legion Aug~ste, 85; Les Unites auxiliaires, 14-
 I I2 Husbands and Wives in the Frontier-Zone
Husbands and Wives in the Frontier-Zone I I3
have resulted in a change in the number of epitaphs erected. Local
seems, mainly, perhaps almost exclusively, ab out the Roman and
elites who were awarded the Roman citizenship are likely to have
Romanized populations. 36
adopted Roman ways of thought and expression. And Italians
Epitaphs may be expected also to provide information only about
sometimes took up residence in the region (though their numbers
that part of the stone-using population which both could afford a
espe~ial1~ in t~e frontier-zones, seem never to have been very large: permanent memorial,37 and was thought, or consldered Itself, to
and ImmlgratlOn appears to have become decidedly less popular
deserve one. 38 It has been said of Roman Bntam that the habIt of
after about AD 100)34 It is at least as likely, however, that any
commemoration 'went fairly far down in the social scale'. 39 It IS
mcrease m the number of epitaphs erected in north Africa was a
likely enough, l suppose, that in most areas 'memorial stones were
consequence of indigenous adoption of the chiefly Roman habit. It
within the reach ofmodest men'.40 But it would be foohsh to thmk
may be, too, that a larger part of the population was able to afford
the cost of a funeral monument. that the poor are not massively under-represented in the epigraphic
record; theyare, as MacMullen has remarked, 'hardly likely to have
l am not denying that there may be a connection between the
left their records upon stone'4' Tbe rank-and-file soldIers sta-
extension of the Roman citizenship and what seems to have been
tioned at Lambaesis, for example, were seven times less likely
an increase in the number of epitaphs erected in north Africa in the
tban their commanding officers to have an inscribed tombstone. 4'
second century AD. It may even be true that the epigraphic habit
Other biases in the record may be suspected. It is not unlikely,
subsided in the third century35 But none of this can be demon-
for example, that the incidence of commemoration varied locally
strated by the charting of data that is really little more than
according to the availability of suitable stone43 So the southern
informed guesswork. All that can safely be said about the
piedmont of the Algerian frontier-zone is under-represented m the
distribution over time of the epitaphs at Theueste (and at many
epigraphic record probably because the Aur"s mountains are the
other north African sites), and really all that Lassere hirnself
most southerly source of large blocks of workable stone 44 What
maintained, is that most of those that have been recovered date
lan Morris has called 'ritual selection' may have mfluenced the
to the period AD 100-300, and that a majority of them may belong
to the second century. habit of commemoration in ways that are not now easIly con-
trolled; the clearest example is the over-representation of ex-slaves
It must be admitted, too, that, for a number of reasons, epi-
on the epitaphs of the city of Rome in the period between about
taphs may mislead as to the incidence (absolute or relative) of a
roo Be and AD roo. 45 Morris maintains also that epItaphs cannot
particular kind of activity or behaviour, either at a specific point
In time or even over the whole of the Roman era. For one thing,
3 Ibid. 206. On epitaphs as a guide to beh~viour, see, a.ls? 1. A.quile.lla. Almer !lnd
6
they tell us, as John Mann has put it, only about people 'who
M. A. Lopez Cerda, 'Determination de la representatlVlte des mscnptt?ns, latmes
used stone inscriptions', in the case of Roman north Africa, it gräce a la statistique inferentielle', Ant. afr. 9 (1975), 115-26; Mache, Urban
Munificence', 182. . ,
37 Roxan, 'Observations', 278 n. 23; 'Wornen on the.Frontters, 462.
3 Morris, Death~Ritual, 158.
8
39 Salway, Frontler People.' 17. . .
4° R. P. Saller and B. D. Shaw, 'Tornbstones and Roman Family RelatlOns m the
34 See P. Ga:nsey, 'Rome's African Empire under the Principate', in P. Garnsey
and C. R. Whlttake!-" (eds), Imperialism in the Ancient World (Cambridge, 197 8), Principate: Civilians, Soldiers and Slaves', JR.S 74 (1984!, 128.. .
1
24 . Cf.~. Romanelh, Slona ~elle pr.ovince romane dell'Africa (Rome, 1959), 109; L.
8 4 MacMullen: Changes in the Roman Emplre: Essays m the Ordmary (Pnnceton,

Teutsch, Gab es Doppelgememden 1m römischen Afrika?' Revue internationale de~ 1990),45· 'b . , 66
droits de l'antiquite, 8 (1961), 281-356; Das Städtewesen in Nordafrika in der Zei't
2
4 Saller and Shaw, 'Tornbstones' , 140 n. 63. See also Roxan, 0 servatlOns, 2 .-

von c.. Gr~cch~s bis zum Tode des Kaisers Augustus (BerIin, 1962). Immigration to 43 Mann, 'Epigraphic Consciousness', 204; A. Mocsy, Gesellschaft und Romant~

the Tnpohtalllan frontier~zone is discussed in D. 1. Mattingly, Tripolitania (Ann sation in der römischen Provinz Moesia Superior (Amsterdam.' 1970!,.166-7.
Arbor, 1994), 143. 44 E. W B. Fentress, Numidia and the Roman Army: Soc/al, Mtlltary and Eco~

35 C~u~es .might be c?njectured: impoverishment? a shortage of suitable stone?

nomic Aspects of the Frontier Zone (Oxford, 1979), 143. .
the natlVlzatIOn of provmcial society? On stone, see 1. C. Mann, 'Epigraphic Con~ 45 Morris: Death~Ritual, esp. at 166. On the over~representatlOn of ex~.slav:s, see
sciousness', JRS 75 (1985), 2°4-6. also G. Woolf, 'Monumental Writing and the Expansion of Roman Soclety m the
Early Empire', JRS 86 (1996), 35-6.
 114 Husbands and Wives in the Frontier-Zone
be expected to describe social practice because they 'were created
for funeral ntuals, and are really telling us about ritual struc-
6
tures'.4 I, too, might wish that more of the texts were not dis-
socmted from their physical context, and that we might somehow
reconstruct the various historical (not just ritual) processes that
were apart of their creation. No one, I think, would claim that
patterns of behaviour in any part of the Roman world are
descnbed perfectly by its surviving epitaphs. It may be said of
alm ost any past society that its dead were treated collectively
accordmg to rules that are now largely unknowable; what is left
behmd must always be understood to be an 'artificial set' 47
Tbe frontier-zone epitaphs of Roman-era Algeria that record
marnages are mostly of two types: those on which a woman
commemorates or is commemorated by a man who is identified
as her husband (maritus, coniunx, ete.); and those on which a man
commemorates or is commemorated by a woman who is identified
as hIs wlfe (uxor, eoniunx, etc.). Together they account for 81.5 per
cent (591/725) of the surviving frontier-zone marriage-epitaphs.
Two other types of epitaphs can be understood to record husbands
and Wlves, and are therefore included among the epitaphs exam-
med below: those on which a man and woman are identified as
'father' and 'mother' (pater ~nd mater, less often, parentes; they
are 8·3 per cent of the fron her-zone marriage-epitaphs, 60/7 25);
and those hke CIL 8. 3°70 (Lambaesis)-D(is) . M( anibus) .
S(aerum) / CaellO' Ma / eedoni' Vet(erano) / Vix(it) . An(nis) .
LX I Feelt . Cassl / a . Coneessa ('To the gods of the underworld:
CassJa Concessa made [this] for Caelius Macedo veteran' he lived
60 years')-on which a woman commemorates' or is co~memo­
rated by a man without indication of their relationship, where the
man and woman have a different nomen (and are therefore unlikely
to be father and daughter or sister and brother), and where the age
of the deceased, If recorded, is said to have been at least 15 (10.2
per cent of the fronher-zone marriage-epitaphs, 74/725).48 . antiken Sozialgeschichte: Festschrift F. Vittingho.tf (Cologne, T9 80),. 243-4·
4
6
Death-Ritual, I6I.
47 Se,e especially 1. Brown, 'On Mortuary Analysis-with Special Reference to the
Saxe-Bmford Re~earch Program', in L. A. Beck (ed.), Regional Approaches to
M~[tuary AnalY~ls (~e~ York a~d London, 1995), 17.
~wo other mscnptIons. are lllcluded among the marriage-epitaphs which are
exammed bel~w: Olle on WhlCh an ex-slave woman is commemorated by her patron
(eIL 8. 1925 -ILAlg I. 3228; Theueste); another on which a slave man and woman
are commemorated by their owner (CIL 8. 2820; Lambaesis).
Husbands and Wives in the Frontier-Zone 11 5
It is generally impossible now to distinguish between Roman
citizens and non-Roman citizens on inscriptions. The only really
certain indications ofpossession of the citizenship are membership
of a Roman tribe,49 and Roman-style filiation (for example, CIL 8.
32 51, Lambaesis: C(aius)' Valerius I C(ai)' F(ilius)' Papiria I
Castus)5 0 Unfortunately, the practice of recordmg tnbal n;ember-
ship and filiation, while fairly common on north Afncan mscnp-
tions of the first century AD, seems almost to have dlsappeared m the
second5' It is sometimes suggested that pos session of the Roman
citizenship is indicated also by Roman-style names, and especially
by the so-called tria nomina (praenomen, nomen, and cognomen. ) 5'
53
But Roman names could be, and were, usurped At least some of
those attested on inscriptions with Roman-style names will have
been Junian Latins (improperly or informally manumitted slaves)54
49 P. A. Brunt, Italian ManpoIVer 225 B,G.-A,D, 14 (Oxford, 1~7~); 208; Le
Bohec, La Troisieme Legion Auguste, 528; R. Taubenschl~g, Clt~z~ns .and
Non-Citizens in the Papyri', in Scritti di diritto romano in onore dl C. Femm (Milan,
1948), iii. 168. . ., .
50 A. M'Charek, Aspects de l'evolution demograplllque et socwle a Mactans aux
Ile et fIle siecles apo J C. (Tunis, 1982),43·
5' See Le Bohec, La Troisieme Legion Auguste, 54· . ,
52 e.g. by H.-G. Pflaum, 'Remarques sur l'ono~astique de Castell~.m Ce~tJan~n~,
in E. Swoboda et al. (eds.), Carnuntina: Ergebnzsse der Forschung uber dIe Glen.,-
provinzen des römischen Reiches: Vorträge beim internationalen Kongress de: Alter-
tumsforscher Carnuntum 1955 (Graz and Cologne, 1956), 134; 'O~omast1~ue de
Cirta', in R. Laur-Belart (ed.), Limes-Studien: Vorträge des 3 I11tern,atlOnalen
Limeskongresses in RheinfeIden/Basel (Basel, 1959), T13; 'Remarques sur 1 ?nom~s­
tique de Castellum Tidditanorum', BCTH (1974-5),25; 1. M. Reynol~s, Inscnp-
tions in the Pre-Desert of Tripolitania', in D. 1. Buck and D. 1. Mattmgly (eds.),
Town and Country in Roman Tripolitania: Papers in Honour of Olwen Ha~kett
(Oxford, 1985),24; see also Meyer, 'Epigraphic Habit', 88 and n. ?9; M .. ~pe1del,
'The Soldiers' Hornes', in Eck and Wolff (eds.), Heer und fntegratlOnspolmk, 47 8.
Cf. Le Bohec, La Troisieme Legion Auguste, 54; MacMullen, Chang~s, 6?; A.
Mocsy, 'Das Namensverbot des Kaisers Claudius (Suet. Claud. 25, 3), KIlO, 52
(197°), 28 7--94; 1. Morris, 'Changing Fashions in Roman Nomenclature m the Early
Empire', Listy Filologicke, I I (1963),46; H. Wo~f~, 'Zum Er~ennt~i~w~rt von ~amen­
statistiken für die römische Bürgerrechtspobtlk der Ka1serzelt, m Studien zur
53 Claudius is said to bave forbidden non-Romans to adopt Lahn na~e~ .(Su:-
I tonius, Claud. 25. 3). Brunt, ftalian ManpOiFer, 708; remarks that th~ pro?1bttJOn 1S
.~ 'better evidence ofthe practice than of its cessatJOn . See also G. Alfoldy, Notes sur
la relation entre le droit de cite et la nomenc1ature dans l'empire romain', Latomus,
25 (1966), 37-57; E. Seidl, Rechtsgeschichte Ägyptens als römische Provinz (Sankt
Augustin, 1973), 130. . . , .
54 Cf. C. R. Whittaker, 'Land and Labour 10 North Afnca, KIlO, 60 (1978),343 n.
65. There were probably a great many of them in the Roman world: P. R: C. Weaver,
'Where Have All the Junian Latins Gone?: Nomenc1ature and Status 10 the Early
 1I6 Husbands and Wives in the Frontier-Zone
For the same reasons, pos session oftwo Roman-style names (nomen
and cognomen, for example) cannot be taken to indicate possession
of the citizenship.
Conversely, it cannat be assumed that a man or woman recorded
with one or more non-Roman names was not a Roman citizen
because some provincials who possessed the citizenship chose no;
to take Roman-style names 55 It cannot be assumed either that
those who are recorded with just a nomen or cognomen were not
Roman citizens, for they may have had other names which have not
been recorded, in some cases, perhaps, as a way of conserving
space on the stone. 56 Praenomina, for example, were rarely
recorded on north African inscriptions after about the time of
Trajan 57
It was remarked earlier that it is normally impossible also to
distinguish between Roman (or Italian) immigrants and Roma-
nized indigenes on inscriptions. The solution that I have adopted
is to divide those attested on the marriage-epitaphs instead into
two groups: those who are demonstrably un-Romanized; and those
who may be categorized either as Roman citizens or as Romanized
indigenes. Several criteria can be used to distinguish the un-
Romanized from the Roman(ized). Auxiliary soldiers,5 8 and those
attested with one or more un-Latinized north African names
(Mababme, for example), or with a single Latinized name of African
origin (Muthunus, for example), can be categorized as un-
Romanized. 59 Among the Roman(ized), we may locate Roman
Empire', Chiron, 20 (1990), 275-305. For JUllian Latins with Roman-style names
see e.g. Pliny, Ep. 10. I04.
55 See Garnsey, 'Rome's African Empire', 250 (on Thubursicu Numidarurn).
56 Cf. M'Charek, Aspects de !'evolution, 44: 'ces elements trahissent sans doute
une condition sodale modeste et peut-etre aussi une condition de pen'!grins.'
57 Le Bohec, La Troisihne Legion Auguste, 54.
58 S~e 1.. F. Giiiiam, 'Romanization of the Greek East: The Role of the Army',
Buj:etm o} the American Society 0/ Papyro!ogists, 2 (1965), 66-7.
Mab~bme: eIL 8. 3081 (:::: 18301). Muthunus: eIL 8. II250. Other examples of
north Afncan names (e.g. Dabar, Guddem, Izelta), and of Latinized names of
African origin (e.g. Baricio, Giddaeus, Zabullus), are collected in M'Charek
Aspects de !'evolution, 93, 101, 154, 184, 186; Pflaum, 'Onomastique de Cirta<
II8; T~ompson, 'Observations', 48-9, 53, 57. The geographical distribution of
the Afncan names attested in Roman-era Algeria is described in E. Frezouls 'Les
Survivances i.ndigenes dans l'onomastique africaine', in A. Mastillo (ed.), L'A/rica
Romana: Altl de! VII convegno di studio, Sassari (Sassari, 1990), 165; in the frontier-
zone, they are most commonly attested at Lambaesis, Thibilis, Tiddis, Thamugadi,
and Verecunda.
Husbands and Wives in the Frontier-Zone I I7
officials, those who belonged to a Roman tribe, legionary soldiers
and officers,60 auxiliary and legionary veterans, and those attested
with one or more Latin names, or with two or more La~l~lzed
names of African origin (Timinius Rogatus, for example). The
union of Marau and Aufidia Silvana at Maktar (CIL 8. 23442) can
be considered to be an example of the intermarriage of Roma-
n(ized) and un-Romanized: Marau' Chubudis' F(ilius) / V(ixit) .
A(nnis) . LXXXII· H(ic) . S(itus) . E(st) / Aufidw . Sllvana .
V(ixit) / An(n)is CHere lies Marau, son ofChubud, [who]lrved 82
years; Aufidia Silvana lived [?] years,)62 .
The methods are crude, and no doubt imperfect. For one thmg,
they cannot adequately describe the partially Roma11lzed. It IS
certainly not the case that all auxiliary soldters were un-Romamzed
right up to the very moment that they were discharged. And there
is no way to distinguish between Roman clt1zen-women and l~dl­
genes who may have taken Latin-style names when they marned.
The scheme that I have proposed 111Ight also be expected to under-
report the frequency of intermarriage, for if any Rom~ns are lrkely
to have quit the chiefly Roman habt! of commemoratmg the dead
on stone and therefore to have gone unrecorded, t! IS precisely
those who married un-Romanized north Africans and parented
'half-African' children. And because the practice of setting up
epitaphs was mainly a Roman custom, they cannot be expected
to describe the marriage habits ofthe un-Romamzed populatlOn-
many north Africans who married other north Afncans Will have
gone unrecorded. ,
' It can hardly be said then that the epitaphs of the Algenan
fron tier-zone are a complete or perfect record of Its marnage-
patterns. What really matters, however, is not whether the figures
they generate are precisely accurate (I am reasonably certam that
they are not), but whether they can serve at least to Identlfy orders
of magnitude from which some valrd concluslOns mlght be
drawn. 63 In what follows, the methods I have descnbed above
60 Though non-Romans are known to have s~rved.in ~he legions ~rom at least t~~
end ofthe Ist cent. AD, the large majority ofleglOnanes m every p~nod ,:",ere Ro~a
citizens' see G. Forni, 11 reclutamento delle legioni da Augusto a DlOclezwno (MI1~n,
, 61 Timinius Rogatus: eIL 8. 3708 (Lambaesls).
1953 ) , 103-5· . . ,
62 Cf. M'Charek, Aspects de !'evolution, !07: 'un example de ma~Ia~e ml~te d un
pereg[r]in d'origine numide avec une romame dont les noms sont ltahqucs ..
63 A similar point is made in R. P. SaUer, Patriarchy, Property and Death m the
Roman Family (Cambridge, 1994), 10-II.
 Husbands and Wives in the Frontier-Zone "9
are used to estimate the incidence of cross-cultural marriage first in
the frontier-zone (defined below); second, at Lambaesis; then at
Theueste and Auzia, the two fron tier-zone sites, besides Lambae-
\
,, sis, where more than fifty marriage-epitaphs have been recovered;
I and lastly, for the sake of comparison, at Thubursicu Numidarum,
,-" ... - ... , , which lay well to the north of the frontier-zone
64

MARRIAGE-PATTERNS

The Frontier-Zone
The frontier-zone is defined here as the region that lay to the south
of a line drawn from Numerus Syrorum east along the Maureta-
nian frontier-road to Auzia, then east and south to Theueste (see
Fig. 2.1). A total of 3,436 epitaphs have been recovered from the
region; 2,624 (76-4 per cent) are intact (or nearly so)6 5 There are
7 2 9 marriages recorded on 725 epitaphs (they are reproduced in
Appendix 1),66 186 (25.7 per cent) of which are mili tary6? The
eighty-eight fron tier-zone sites where the marriage-epitaphs have
been recovered are identified in Fig. 4-4; the number found at each
site is indicated in parentheses in the accompanying legend. Table
4. land Fig. 4.5 indicate the status of the husbands and wives
attested; they are categorized as Roman(ized) or un-Romanized
according to the criteria described earlier. Twenty-two epitaphs
which record the 'marriages' of slaves Of of non-Romans who 68
were not of African origin (Greeks, for example) are excluded
Of the 186 Roman(ized) soldiers and veterans attested, just one

64 On the marriage-patterns attested at Lambaesis and Thubursicu Numidarum,

see also D. Cherry, 'Marriage and Acculturation in Roman Algeria', Classical
Philology, 92 (I997), 7 1- 8 3.
65 The figures count singly the many epitaphs that are published both in eIL 8
and in ILAlg:r; on duplication, see Meyer, 'Epigraphic Habit', 82 ll. 43·
66 Two marriages are recorded on each offour epitaphs: eIL 8. 3296, 9053, 90 65,
9 116 . I have included among the marriage-epitaphs fOUf which attest betrothai (CIL
8.2857,3065,3485,4318), and one which records concubinage (CIL 8. 9 100).
67 Epitaphs on which the deceased or the commemorator is identified as a soldier
or veteran have been classified as 'military'. The attested rate of 'marriage' in the
military population of the frontier-zone is 30.8% (186/6 04).
68 eIL 8.189 8 (= ILAlg I. 3135). 1899 (= ILAlg I. 3140), 1931 (= ILAlg I. 3244),
2255, 2803a, 2820, 3290, 3292, 3463, 3521,3563, 3597, 3930, 3935,4046,4071,4152,
437 2-3, 10628 (= ILAlg I. 3139), 16561 (= ILAlg I. 3134), 18392; Leschi, Etudes
d'epigraphie, 185.
 120
Husbands and Wives in the Frontier-Zone Husbands and Wives in the Frontier-Zone 121
Table 4· I. Husbands and wives on the fron tier-zone epitaphs
Civilian population: intermarri.age of
Military population: intermarri.age of Roman(ized) and un-Romal11zed
Roman(ized) Un-Romanized TOTAL Roman(ized) and un-Romal11zed ~ / (1.7%)
husbands husbands (0.3%)

Military Civilian Military Civilian

Roman(ized) wives 185 I

Un-Romanized wives I
5
o I Military population:
TOTAL 186 Roman(ized)
6
husband and wife
(26.2%)
(0·5 per cent), a legionary soldier (mi/es) named C. Harnius
Maccus, appears to have married an un-Romanized north African
woman (Mababme).69 Much the same can be said of the civilian
population. Ofthe 512 Roman(ized) Wornen attested, only five (1.0 Civilian
population: un-
per cent) appear to have married un-Romanized north Africans: Romanized
Iulia Fortunata, who was commemorated by her husband, Hard- husband and wife
alius; Hariana Rufilla, who was cornmemorated together with her (0.1%)
sou's father, Themarsa; Aelia Fortunata, who was commemorated
by her husband, Narnpamo; Victoria, who was comrnemorated by
her husband, Numidi; and Sextilia Spica, who commemorated
her husband, Muthune70 Similarly, just seven of the 5 1 4 Roman- N = 707
(ized) men (1.4 per cent) appear to have married un-Romanized
women: Aelius Maarnon, who commemorated his wife, Thereba Fig. 4.5. /ntermarriage in the Jrontier-zone
(perhaps Therefna); C. Iulius Victor, commemorated by his wife,
Baricas; Ulpianus, commemorated by his wife, Siddina; Plotius Saturninus, commemorated by his wif~, ~Icoforoi; and Aemilius
Pequarius, commemorated by his wife, Mustel; Felix, who was Niger, cornrnemorated by his wife, Bhang. Of the 7 Roman-
commernorated together with his wife, Tsedden; C. Cornelius Th data may be expressed anot er way. . 00 I
C d)e rnen attested on the military and civilian epllaphs, on ~
69 Mababme: eIL 8. 3081 (= 1830I; Larnbaesis). The only auxiliary soldier
eli~~t (LI per cent) appear to have married un-Romamzed nort
attested, Abillahas, who served with the second cohort Sardorum at Rapidum
(CIL 8. 919 8 ), seems to have been married to a Roman(ized) woman (Sextia Prima). 1 - ILAI 1 3747' EI ma el Abi6d); the nam~
For the 185 soldiers and veterans with Roman(ized) wives, see Appendix 3 (i, a). 7 Aeliu~ Maamon: eIL 8. 207 8 ( - C. ~ulius Vi'etor: eIL 8. 4501 (~r. Si~l
7° Iulia Fortunata: CIL 8. 1954 (= 16513 = ILAIg I. 3148; Theueste); on the name Therefna IS attested at ILAlg J. I?I3. e Pflaum 'Castellum Celtlanum,
Hardalius, see CIL 8, p. 225 (aeeording to Orosius 7. 36, Ardalio was the name of a Khallef); for Baricas as a n~rth Afnc~n n,~~, ~; ('Baric'). Ulpianus: eIL 8. 9077
river in the region between Theueste and Ammaedara). Hariana Rufilla: CIL 8. 137 ('Barieus'); Thomp~on: Obser~~~~nsr' .'224. Plotius Pequarius: eIL ,8.9 1 52
251I~12 (Caleeus Hereulis). Aelia Fortunata: CIL 8. 3347 (Lambaesis); for (Auzia); on the name Slddm~, see pfl. g ,'tastellum Tidditanorum', 27 ( Muste-
Nampamo as a north Afriean name, see Pflaum, 'Castellum Celtianum', 137; (Auzia); for the ~ame,Mu.ste, see . a~mEtudes epigraphiques (Paris, 197 8), 1 83
'Onomastique de Cirta', II8; 'Castellum Tidditanorum', 27. Vietoria: CIL 8. eus', 'Musteolus); L Afnque ro.maz?e. 'Mustus'). Felix: ILAlg I. 2947 (Aquae
4 1 51 (Lambaesis); on the name Numidi, see 1. Kajanto, The Latin Cognomina ('Musti-'); Thompson, 'Observatlon~i17 ( p 279 C Cornelius Saturninus: ILAlg
(Helsinki, 19 6 5), 206 ('Numida'). Sextilia Spiea: eIL 8. 17702 (Maseula); on the Caesaris); on the .name T~e~den,. see. M -?z~t ;Insc~iptions inMites de l' Aures:' I 5~
I. 37 4 (Hr. el GIS). Aemlhus ~lger. h on
name Muthune, see Thompson, 'Observations', 50, 57 ('Muthun'). The 507 epitaphs 8 'Barig see Thompson, 'ObservatIOns,
(no. 8; Valley of the Oued Abdl); on t e name ,
that record Roman(ized) wives and husbands are identified in Appendix 3 Ci, b).
48, 57 ('Barie').
 122 Husbands and Wives in the Frontier-Zone
Africans. And just six of the 698 Roman(ized) women (0.9 per
cent) had an un-Romanized husband.
lt may be said again that the epitaphs are not aperfect record of
the marriage-patterns of the frontier-zone population. They prob-
ably under-report the frequency of intermarriage: it may be
doubted that only I per cent (14/r,398) of all the Roman(ized)
men and wornen in the frontier-zone married un-Romanized north
Africans. The figures can therefore be taken to indicate only some
very rough orders of magnitude. But it may be remarked also that
they would have to be wrong by a factor of about 100 before it
could be maintained that a majority even of the soldiers and
veterans married un-Romanized wornen.
Lambaesis
Lambaesis was horne to the Third Augustan legion probably from
about AD 115/r7 until 238, when the legion was temporarily dis-
banded (perhaps for twenty years)7' Three camps in all were con-
structed: the first, the so-called 'camp de l'est', seems to have been
built in AD 81;13 another, the 'camp de l'ouest', was constructed
probably between 81 and 129; the largest of the three, the 'Grand'
camp, was completed in 129.74 The civilian settlement that grew up
around the camps was given Latin rights sometime between AD 15 8
and 16 I, styled a municipium at least by tbe time of Caracalla, and
made a colony (colonia) probably shortly after 238.75
The camps and civilian settlement at Lambaesis together have
yielded alm ost 1,400 intact epitaphs76 Several cemeteries have
72 In ~eI?-eral, see H. D'Escurac Doisy, 'Lambese et les veterans de la Iegio tertia
Augusta, In~. ~enard (ed.), Hommages ci Albert Grenier (Brussels, 1962), 57 1- 83;
Fentress, Numldw, 94-6; M. Jano~, 'Recherehes a Lambese', An!. air. 7 (1973),193-
254, and 21 (1985), 35-102; there 18 a good map ofthe region in Janon (at 253). AD
IISIIT Le Bohec, La Troisieme Legion Auguste, 36 2.
• 73 For the date, see especially Fevrier, Approches du Maghreb romain, i. 111. There
IS a plan of the camp in Le Bohec, La Troisieme Legion Auguste, 363.
74 See Janon, 'Rech~~~hes a ,Lambese', 211-12. For a plan of the 'Grand' camp,
see .Le Bohec, La TrOisleme Legion Auguste, 41 I.
7:> Latin !'ights: CIL 8. 18218 = ILS 6848. Municipium: CIL 8. 18247; see also P.
MacKendnck, The North African Stones Speak (Chapel HiIl, NC, 1980), 221.
Colony: see Broughton, Romanization 0/ A/rica, 138; cf. H.-G. Pflaum, 'La Roma-
nisation de l'Afrique', Vestigia, 17 (1973), 64 (AD 210).
76 On the epitaphs, see especially Lassere, 'Recherches sur la chronologie', 9 -
6
I07·
Husbands and Wives in the Frontier-Zone 12 3
been identified: one (I) was alongside the road that ran north
from the 'Grand' camp; another (11) was located to the east of
the 'Grand' camp, between the Oueds Necheb and Markouna; a
third (111) was situated about 500-700 m. to the west of the
forum near the Oued Tazzoult; what has sometimes been taken
to be a fourth cemetery (IV), located south-west of the 'Grand'
camp, may bave been merely a continuation of III (it has been
entirely destroyed).77 .
About a quarter ofthe intact epitaphs (380) record marnages. A
total of 762 husbands and wives are attested (eIL 8. 3296 records
two marriages), induding 140 soldiers and veterans (36.7 per cent
of the 381 husbands). Table 4.2 and Fig. 4.6 describe tbeir status,
which is categorized as Roman(ized) or un-Romanized; fifteen
epitaphs which record the 'marriages' of slaves or of non-Romans
who were not of north African origin are excIuded. 78
Just one of the 364 Roman(ized) men recorded on the military
and civilian epitapbs (0.3 per cent), the legionary, C. Harnius
Maccus, seems to have married an un-Romanized wornan
(Mababme)79 And only two of the 365 Roman(iz~d) women
recorded (0.5 per cent) appear to have been marned to un-
Romanized north Africans: Aelia Fortunata, who was commemo-
rated by ber husband, Nampamo; and Victoria, commemorated
Table 4.2. Husbands and wives on the epitaphs 0/ Lambaesis
Roman(ized) Un-Romanized TOTAL
husbands husbands
Military Civilian Military Civilian
Roman(ized) wives 139 224 o 2
Un-Romanized wives o o o
TOTAL 140 224 o 2
77 Le Bohec, La Troisieme Legion Auguste, 107-8 (with map); see also Leschi,
Etudes d'epigraphie, 36-5I.
7 CIL 8. 2803a, 2820, 3290, 3292, 3463, 3521, 3563, 3597, 3930, 3935, 4046, 4°7 1,
S
4152, 18392; Leschi, Etudes d'epigraphie, 185. .
79 C. Harnius Maccus: CIL 8.3081 (= 18301). For the 139 soldlers and veterans
who had Roman(ized) wives, see Appendix 3 (ii, a).
 12 4 Husbands and Wives in the Frontier-Zone Husbands and Wives in the Frontier-Zone 12 5

Military population: intermarriage of

Roman(ized) alld un-Romanized (0.3%)
has recently dated its promotion instead to the time of Marcus
Aurelius, on the grounds, first, that it would shorten the interval
between its promotion and the earliest recorded mention of it on
a surviving inscription, ILAlg I. 3°32, which belongs to the time
of Commodus, and second, that the legion was at Lambaesis

I already by AD 8 I 8, It is indeed unlikely that Theueste's promo-

tion went unrecorded for more than half a century. But an

I inscription cannot be assumed never to have existed because it

has not been found. And Michel Janon has demonstrated that
Military population:
Roman(ized)
husband and wife
(38.0%)
I the building activity attested at Lambaesis in AD 81 ('le camp de
1'est') was not connected with the transferring of the whole
legion. 83
A total of 215 intact epitaphs have been recovered at Theueste.
About a quarter of them (55) record marriages, including those of
four soldiers (7.3 per cent of the husbands). The husbands and
wives who are attested are categorized as Roman(ized) or un-
Romanized in Table 4.3 and Fig. 4.7; five epitaphs which record
the 'marriages' of slaves or of non-Romans who were not of north
African origin are excluded. 84
All four of the military husbands attested (two legionaries and
two veterans) appear to have had Roman(ized) wives85 And only
Civilian population:
intermarriage of one of the fifty Roman(ized) women (2.0 per cent) seems to have
N~ 366 Roman(ized) alld Ull- married an un-Romanized uorth African: Iulia Fortunata, who
Romanized (0.5%)

Fig. 4.6. Intermarriage at Lambaesis

Table 4.3. Husbands and wives on the epitaphs 0/ Theueste
80
by her husband, Numidi. The incidence of intermarriage attested
across the whole of the population is just 0.8 per cent (3/3 66). Roman(ized) U n - Romanized TOTAL
husbands husbands

Theueste Military Civilian Military Civilian

Horne to the Third Augustan legion from AD 75 until about I 15/ 17, Roman(ized) wives 4 45 o 50
when the leglOn was moved to Lambaesis, Theueste (mod. Tebessa) U n -Romanized wives 0 o o o o
was made a colony probably shortly after the legion left 8 ' Meyer TOTAL 4 45 o 50

80 Aelia Fort~nata: C:1L 8. 3347. Victoria: eIL 8. 4151. The civilian epitaphs that
re~?rd RO~an(lz:d). WIV S and hus~ands ar~ listed in Appendix 3 (il, b).
7
AD 75- C. Damels, The Frontlers: Afnca', in 1. S. Wacher (ed.), The Roman R2 'EpigraphicHabit', 83-5 andn. 49. 83 Janon: 'Recherchesa Lambese', 21 2.
World(Lo,ndon and N~w York, 1987), i. 240. Colony: ibid. i. 242. The exact Ioeation 8, eIL 8. 1898 (~ILAlg I. 3135), 1899 (~ILAlg I. JI40), 1931 (~ILAlg I. 3244),
of th,e leglOnary base 18 unknown: V. A. Maxfield, 'The Frontiers of the Roman 10628 (~ ILAlg I. 3139), 16561 (~ ILAlg I. 3134).
Emplre: Same Recent Work', JRA 2 (1989), 343. 85 eIL 8. 16544 (:= ILAlg 1. 3I06), 16545 (= ILAlg I. 3107); ILAlg r. 3105, 3121.
 126 Husbands and Wives in the Frontier-Zone Husbands and Wives in the Frontier-Zone 127

Civilian population: Table 4-4. Husbands and wives on the epitaphs 0/ Auzia
intermarriage cf
Rornan(ized) and un-
Military population: Roman(ized) Un-Romanized TOTAL
Romanized (2.0%) husbands husbands
Roman(ized) husband and
wife (8.0%)
Military Civilian Military Civilian

Roman(ized) wives 6 o o 75
Un-Romanized wives 0 o o 2

TOTAL 6 o o 77

town (municipium) was later established at the site; it was made a

N~50
Fig 4.7. Intermarriage at Theueste
was commemorated by her husband, Hardalius (named probably
for a nearby river),86
Auzia
In AD 24, according to Tacitus (Annals 4, 25), Tacfarinas' Musu-
lamian warriors, having been forced to abandon tbeir siege of
Thubursicu Numidarum, set up camp near the 'half-ruined' (sem i-
rutum) native fort (castellum) at Auzia (mod, Souk EI Ghoziane),
which they themselves had burnt earlier, and where they were
subsequently routed by the proconsular governor, P Cornelius
Dolabella (Tacfarinas hirnself fell in the fighting), A Roman-style
86 Iulia Fortunata: eIL 8. 1954 (= r6513 :::: ILAlg
3148). For the 45 civilian
I.
epitaphs that record Roman(ized) wives and husbands, see Appendix 3 (iii, b).
colony probably in the time of Septimius Severus87
The sile has yielded 133 intact epitaphs, most ofwhich appear to
date to the period AD 225-5088 Seventy-fonr epitaphs record
seventy-seven marriages, including those of six soldiers and veter-
ans,89 Table 4A and Fig, 4,8 describe the statns-Roman(ized) or
un-Romanized-of the 154 husbands and wives who are attested,
The rate of intermarriage that is indicated, 2,6 per cent (2/77), is
almost the same as at Theueste, Seventy-five of the seventy-seven
Roman(ized) men (including the six soldiers and veterans) appear
to have had Roman(ized) wives; two others seem to have married
un-Romanized north African women; Ulpianus, who was comme-
morated by his wife, Siddina; and Plotius Pequarius, who was
commemorated by his wife, MustePO
It might be supposed that inadequacies in the methods I have
used to distinguish Roman(ized) and un-Romanized explain why
intermarriage seems to have been so uncommon in the fron tier-
zone, It is unlikely, however, that methodological shortcomings
alone can account for the patterns the epitaphs describe, for
when the very same methods are used to categorize the marriages
attested on the epitaphs of Thubnrsicu Numidarum, a significantly
higher rate of intermarriage is indicated,
eIL 8, p, 769.
87
Thirteen epitaphs at Auzia are securely dated (eIL 8. 9065, 9077, 9085: 9 086 ,
88
909°,9091,9109, 9IlI, 9II5, 9II6, 9133, 9158, 9162); aH belong to the penad AD
22'-50.
&"9 Three epitaphs each record two marriages: eIL 8. 9053, 9065, 9 II ?"
9° Ulpianus: eIL 8. 9077. Platius Pequar~us: C:1L ~. 9152. T~e ep~taphs that
record Roman(ized) husbands and wives are Identtfied m AppendIx 3 (IV, a-b).
 128 Husbands and Wives in the Frontier-Zone 129
Husbands and Wives in the Frontier-Zone
Table 4.5. Husbands and wives on (he epitaphs 0/ Thubursicu Numidarum
Civilian
Roman(ized) U n ~ Romanized TOTAL
Military population: husbands
Roman(ized) husbal1d and
husbands
wife (7.8%)
Military Civilian Military Civilian

Roman(ized) wives 3 133 o 9 145

Un-Romanized wives 0 12 o 2 14
TOTAL 3 145 o II 159

N ~ 77
Fig. 4.8. Interman"iage at Auzia
Thubursicu Numidarum
Long a tribaI centre of the Numidae, Thubursicu Numidarum was
made a mumClplum (Ulpium Traianum Augustum Thubursicu) no
later than AD ,11 3: and a colony sometime before 2709' It is Paul
Ma~Kendnck s Vlew that the settlement was 'thoroughly Roman-
Ized ; perhaps as early as the time of Tacfarinas' rebellion (AD 17-
24)· 9 But there are several indications also of a sizeable and
probably only parlIally Romanized, indigenous population: two
Baal-Saturn temples; 173 Punic stelae; Latin inscriptions bristling
wlth north Afncan names.
1
9 Mun~cipium: [LA/g. 1. ,I240; see also Garnsey, 'Rome's African Ern ire' 2
~~1~hS~WGs!tte~!:~~s~~1(~!r~f~~:~~g~~r~rai91~Y~), Hisloria, 30 (1981), ~50;' i;~en~
92 M K d ' k ' .
include~Ct:~lf~~~ :b":~~th African Stones, 216- 17, citing its public buildings, which
date). ' a s, a monumental arch, and a theatre (probably of Severan
In all, 529 epitaphs have been recovered at the site, 159 of whieh
(30.1 per cent) record 162 marriages, including those of three
veterans (1.9 per cent of the 162 husbands).93 Table 4·5 and Fig.
4.9 describe the status-Roman(ized) or un-Romanized-of the
husbands and wives who are attested (three epitaphs which record
the 'marriages' of non-Romans who were not of north African
origin are excluded)94
Of the 148 Roman(ized) men attested on the epitaphs (including
the three veterans), twelve (8. I per cent) appear to have been
married to un- Romanized north Africans. Six were commemo-
rated together with their wives: Iulius Primulus with Namfamina,
C. Iunius Saturninus with Secchun, Paternus with Sesola, Felix
with Sahnam, Helvius Saturninus with Thadir, Iunius Felix with
Namgedde. Five others-Fronto Lepta, Florus, Gallus, Quirinius,
and L. Aemilius Rogatus-were the husbands respectively of Ber-
iet, Zabulla, Berecbal, Therefnat, and Namgedde. Arius Felix
commemorated himself and his wife, Namgidde 95 Of the 145
93 The marriage-epitaphs are reproduced in Appendix 2. Three epitaphs each
record two marriages: eIL 8. 5064 (:=::: ILAlg I. 1810); 5II2 (= ILAlg I. 1590); ILAlg
1. 1976. I have inc1uded one epitaph that records betrothaI (lLAlg J. 1503). Among
the marriage-epitaphs are 90 (56.6%) on which a woman commemorates or is
commemorated by a man who is identified as her husband, or a man commemorates
or is commemorated by a woman identified as his wife; 68 (42.8%) on which a
woman commemorates or is commemorated by a man without indication of their
relationship, where the man and woman have a different nomen, and where the age
ofthe deceased, ifrecorded, is said to have been at least 15; and I (0.6%) on which a
.B man and woman are identified as pater and mater.
94 eIL 8. 4953 (~ILAlg 1. 1839), 4999 (~ILAlg 1. 19°5), 5054 (~ILAlg 1. q81).
95 Namfamina: eIL 8.5°55 (::::: ILAlg J. 1666); for Namfamina as a north African
name, see M'Charek, Aspects de !'evolution, 186 ('Namphamina'). Secchun: eIL 8.
5099 (::::: ILAlg I. 1713). Sesola: eIL 8.5103 (= ILAlg 1. 1805). Sahnam: ILAlg I.
154 1; variations of the name are attested at ILAlg I. 919, 1006 ('Sanam'), 1059
 130 Husbands and Wives in the Frontier-Zone Husbands and Wives in the Frontier-Zone 13 1
Civilian population: un-Romanized
Civilian population:
The data may be summarized this way: fully 13.2 per cent (21/
husband and wife (1.3%)
intermarriage of Roman(ized) 159) ofthe marriage-epitaphs appear to record the union ofRoman-
un-Romanized (13.2%) (ized) and un-Romanized, compared to just 0.8 per cent (3/3 66)
of the marriage-epitaphs recovered at Lambaesis.
Military
population:
Roman(ized)
Conclusion
husband and The marriage-patterns described by the epitaphs are summarized
wife (1.9%)
in Table 4.6. The figures are at best a rough estimate of the
frequency with which Roman(ized) married un-Romanized eitlrer
in the fron tier-zone or at Thubursicu Numidarum. They prob-
ably underestimate the actual rate of intermarriage at every site.
And they misrepresent the marriage habits of the indigenous
population. Twelve of the fourteen un-Romanized women attested
at Thubursicu Numidarum, for example, appear to have had
Roman(ized) husbands; put another way, just two of the 159 mar-
riage-epitaphs seem to record the union of two un-Romanized
north Africans: ILAlg 1. 1634, on which a man named radar is
commemorated together with his wife, Zabullica; and ILAlg 1.
N ~ 159
1774, where a woman named Sumuda is commemorated together
with her husband, Mustiolus. 97 It must be the case that many
Fig. 4·9· Intermarriage al Thubursicu Numidarum marriages of un-Romanized north Africans have not been
Roman(ized) women attested, nine (6,2 per cent) seem to have recorded.
marned un-Romamzed north Africans: three were commemorated There is, however, no reason why the data in Table 4.6 should be
more inaccurate in the case of one community than of another, or
rtogether
N r
wlth thelr husbands-Vivia Matrona wI'th B . h' A .
. anc 10, eml- why the epitapbs at one of the sites should be any more or less
Ja. ata IS wrth Bazabulus, Germana with lamascai; three more-
luha Honorata, Postuma, and Sextilia Villatica-are said to have representative of its population. And there is no evidence to show
been the wlves respectively of Zabo, Numida, and Nabor; three that the un-Romanized population in the frontier-zone was
othe~s-Saturmna, Petronia Frontilla, and lulia Privata-were all
marned to men named Mustiolus.96 Germana: ILAlg I. 1635. Iulia Honorata: CIL 8. 5018 (:::; ILAlg I. 1683); on the
name Zabo, see Thompson, 'Observations', 57 ('Zabullus'). Postuma: CIL 8. 506 9
(= ILAlg I. 1824); for the name Numida, see Kajanto, Latin Cognomina, 206.
}i~h!n~mt'), r~oI ('shahnaim'), 2315 ('Sanamt'). Thadir: ILAlg I. 1616. Namgedde' Sextilia Villatiea: CIL 8. 5107 (= ILAlg 1. 1891); on the name Nabor, see Pflaum,
g . 1703, on t e name Namgedde (and N "dd b 1) , . 'Onomastique de Cirta', I I 8 ('Nabor-'). Saturnina, Petronia Frontilla, and lulia
Aspects de !'evolution 186. Beriet' eIL 8
2
a~gl e, e ow , see M Charek, Privata: CIL 8. 5098 (= ILAlg I. 1876); ILAlg I. 1356; ILAlg I. 1977; on the north
attested also at eIL 8 '62 . 82
. 49 4 (- ILAlg I. 15 ); the name is
Cirta' 118 ('R . t') . Z ~2'I12~499, 255°7, 27713; see also Pflaum, 'Onomastique de Afriean origin of the name Mustiolus, see Pflaum, 'Onomastique de Cirta', II8;
'Puni;o-Lib a ~ne . a u a. CIL 8. 17201 (:::; ILAlg I. 1950); for Zabulla as a 'Castellum Tidditanorum', 27·
ILAlg I. 14;8;nth~a~~es~: a~~~~~s~l~~ ~b~~z~tions', 57 (,Zabullus'). Bereebal: 97 Similarly, just one marriage of un-Romanized north Afrieans is attested in the

I. 1913. Namgedde: CIL 8 4906 (:::; ILAl

2
. 17 93, ~7507. Therefnat ILAlg frontier-zone, that of Amausgaris and Taseuri at Hf. Metkides (CIL 8. 2200 =
the 13 6 men who had R . (' d) . g I. 1396). N~mgldde; ILAlg I. 1417· For ILAlg I. 2975); the other fourteen un-Romanized north Afrieans attested on the
96 .• oman Ize wlves, see Appendix 3 (v, a-b). fron tier-zone epitaphs all appear to have had Roman(ized) partners. Variations of
VlVla Matrona; CIL 8. 5132 (:::; ILAlg I. 1435); for Bariehio as a north Af . ladar are attested at CIL 8. 9923 (ladir), 10686, 12102, 12207, 17253 (lader); for
name, see Thompson, 'Observations', 48 ('Barieio'). Aemilia Natalis: ILAlg I. ~~e3a7~ Zabulliea as a north Afriean name, see Thompson, 'Observations', 57 ('ZabuUus').
 132 Husbands and Wives in the Frontier-Zone Husbands and Wives in the Frontier-Zone 133
Table 4.6. The incidence 0/ interman'iage at selec/ed significant measure of cross-cultural interrelationship in the
sites in Roman-era Algeria
fron tier-zone, still less of what Elizabeth Fentress has called
Site Intermarriage as % of all recorded marriages 'assimilation'.9 8
It is virtually impossible now to identify marriage-patterns
Military Civilian TOTAL within the seemingly self-segregated Roman(ized) populatIOn of
population population the frontier-zone, for almost nothing survives to descnbe the
origins of the wives who are named on the frontier-zone epitaphs.
Auzia 0.0 2.8 2.6 We might suppose that some of them had been drawn to the army-
Lambaesis 0·7 0·9 0.8 towns that were scattered across the region, where they could expect
Theueste 0.0 2.2 2.0 to make a living (of sorts) by providing services-Iaundering,
Frontier-zone I.I 2·3 2.0
Thubursicu Numidarurn entertainment prostitution 99 Some will have been ex-slaves, hke
0.0 13·5 I3· 2 the two wom~n commemorated at Theueste who married their
former owners. IOO
I suspect that a great many of the soldiers' wives were other
significantly smaller than at Thubursicu Numidarum. Compari- soldiers' daughters. TOI There is no evidence to show that soldlers
son, in other words, can be made across different sites even if the or their children married into the families of the Romanized local
data are suspected of misrepresenting historical practice. elites or of the merchants and craftsmen (carpenters, masons,
The marriage-patterns that are recorded on the epitaphs of the metal-workers) who are likely to have taken up residence in the
frontier-zone and Thubursicu Numidarum are compared in Table civilian settlements which grew up around many of the frontler-
4· 7· A standard chi-square 2 X 2 table statistical test applied to zone army-camps. '0' It might be said of the fron tier-zone army in
the data would yield a p-value of less than .001, from which it north Africa as it has been said recently of the soldlers statIOned
could be inferred that there is a very significant difference statis- in northern' Britain, that it constituted its own 'distinct and
tically between the rates of intermarriage that are attested in the separate' society. <03 The marriage-patterns described by the fron·
fron tier-zone and at Thubursicu Numidarum. The figures would tier-zone epitaphs would seem to indicate that there was, m
seem to demonstrate therefore that the intermarriage of Roman- Shaw's wards, a 'terrible estrangement' between the army and
(ized) and un-Romanized was decidedly less common in the
fron tier-zone, perhaps especially at Lambaesis, across both the
military and civilian populations. There is no evidence of any 98 Fentress: Numidia, 78; see also M. Brett and E. W. B. Fentress, The Berbers
(Oxford. 1996), 55· .
99 See especially H. von Petrikovits, 'Lixae', in w. S. Hanson and L. J. F. Kepple
(eds.), Roman Frontier Studies 1979 (Oxford, 1980), 1027-34·
Table 4-7· Marriage-patterns in theJrontier-zone and 100 eIL 8. 1925 (= ILAIg I. 3228), 27869 (::::: ILAIg I. 3199). See also A. 1.

at Thuhursicu Numidarum
Frontier-zone Thubursicu
Numidarum
Marshall 'Roman Warnen and the Provinces', Ancient Sodety, 6 (1975), I~7. "
JOT Th~ idea is borrowed from C. M. WeHs, '''The Daughters of the ~eglment :
TOTAL
Sisters and Wives in the Roman Army', forthcoming in the proceedmgs of the
XVlth Roman Frontier Studies Conference (Aug. I995)· .. . ..
102 On army-towns, see especially C. S. Sommer, The Mlbtmy VICl In Roman

Britain (Oxford, 1984), 3I-4. For merchants, Sal~ust, Bell. Iu~. 4.4., .
Intermarriage of Roman(ized) 10 3 I. Ferris, 'Shopper's Paradise: Consumers m ~oman BntaI,n, m Ru.sh (ed.),

and un-Rornanized 14 (2.0%) 21 (13.2%) Theoretical Roman Archaeology, I36. Cf. C. van Dnel-Murray, Gender m Que.s-
35 tion', in Rush (ed.), Theoretical Roman Archaeology, I2, 16: on the. Dutc~ army m
Roman(ized) husband and wife 693 (98.0%) 138 (86.8%) 83 1
TOTAL Indonesia; K. Ballhatchet, Race, Sex and Class under the Ra}: Imperial At~l!.udes and
70 7 159 866 Policies and their Critics, 1793-1905 (New York, I980), 2, I67, on the Bnttsh army
in India.
 134 Husbands and Wives in the Frontier-Zone
indigenons society, and perhaps more generally between the
Roman(ized) and the nn-Romanized populations. W4
It might be argued, I suppose, that some Roman citizens
who married non-Romans will have wanted to conceal it
because, through much of the period from about the Social
war (c.90 BC) until after the death of Septimius Severus the
intermarriage of Roman and non-Roman was invalid ~nder
Roman law and, in various ways, penalized. There are, however,
a number of reasons for believing that the legal rules had little
practical effect.
LAW AND SOCIAL PRACTICE
It was a long-standing principle in Roman law that a lawful
marriage could be contracted only by two Roman citizens or by
a Roman citizen and a 'foreigner' (peregrinus/a) or Latin who
possessed conubium, which the Roman jurists defined as 'the capa-
cIty to contract a valid marriage'. W5 As early as the Latin settle-
ment of 338 BC, conubium was identified as a right (ius conubii)
distinct from the citizenship. w6 But it seems rarely to have been
awarded without simultaneous enfranchisement. 1°7
Children born to a marriage for which the partners possessed
conubium took their father's status. w8 So the child of a Roman
1°4 Sh~w: 'S.oldiers and So~iety', 144; cr. 148: there are, he conc1udes, 'few signs'
that the IsolatIOn of the soldtery at Lambaesis was 'in auy way mitigated by other
normal müdes of interrelation within civil society (e.g. intermarriage),. See also
MacMullen, Changes, 226. . 61-2), was held to be unlawful (iniustum), but appears not to have been legally vOld;
1°5 Tituli Ulpiani 5.3 (conubium est uxoris lure ducendaefacultas); cf. Gaius Inst.
I: 56. !he Ti~uli Uipi~ni is probably an early 4th cent. AD epitome of a Regularum
b~e~ smgularzs of Ulplan. On conubium, see especially F. de Visscher, 'Conubium et
ClVl~a~', l!-evue internationale des droUs de l'antiquite, 1 (1952), 401-22; 'Conubium
~t ClVlt~S , Iurt!, 7 \1951), 140-4, .is a, s~orter version of the same paper; E. Volterra,
La no~~one gmndlca deI conu~n~m , 111 Studi in memoria di E. Aibertario (Milan,
1953), 11. 34~-84; se~ .also F. Vltt111ghoff, 'Militärdiplome, römische Bürgerrechts~
und Int~gratlO~~poht1k der Hohen Kaiserzeit', in Eck and Wolff (eds.), Heer und
Integ~atl.onspollflk, 543. Other requirements of a valid .marriage are discussed in
Tr;§llan, Roman ~arriaße, 38-43, 54-7.
107 A. N. Sherwm-Whlte; The R?n:~n Citizenship, :nd edn. (Oxford, 1973), 32 .
See also D. Cherry, The MIllIClan Law: Marnage and the Roman Citizen~
ship', Phoenix, 44 (1990), 245-6.
lOS Tituli Uipiani 5. 8: 'where the right of intermarriage exists, children always
fol1ow the father' (conubio interveniente fiberi semper patrem sequuntur); Digest I. 5.
19 (Celsus): 'where a lawfuJ marriage has been contracted, the children follow their
Husbands and Wives in the Frontier-Zone 135
man and a foreigner or Latin who had conubium was born a
Roman citizen. Conversely, the child of a Roman woman and a
10
non-Roman who possessed conubium was born non-Roman. 9
Under the rule of the 'Iaw of nations' (ius gentium), which com-
prised those legal principles that the Roman jurists considered to
be common to all peoples, including Roman citizens, children born
to a marriage for which the partners did not have conubium were
illegitimate and therefore took their mother's, not their father's,
status."o Their condition was thought to be analogous to that of
children conceived outside of marriage. I I I So children born to a
Roman man and a non-Roman woman (foreigner or Latin) who
did not have conubium were illegitimate non-Romans; those born
to a Roman woman and a non-Roman (foreign or Latin) man who
112
did not possess conubium were illegitimate Roman citizens.
The rules that governed the status of children born to marriages
for which the partners did not have conubium were overthrown by
the Minician law. "3 Passed probably sometime before the Social
father' (cum legitimae nuptiae factae sint, patrem liberi sequuntur). The law is
summarized in Gaius, Inst. I. 56.
10
9 Gaius, Inst. I. 77. . .
110 Tituli Ulpiani 5. 8: 'where the right of intermarriage does not eXlst, they [l.e.
children] take their mother's status' (non interveniente conubio matris condicio~i
accedunt); Cicero, Top. 20: in the absence of conubium, 'children do not follow thelr
father' (qui nati sunt patrem non sequuntur); Isidorus, Orig. 9. 7. 21: 'whenever there
is no right of intermarriage, children do not follow their father' (quotiens autem
conubium non est filii patrem non sequuntur).
l ! l Digest I. 5.' 19 (CeJsus). A marriage for which the partners lacked conubium,
such as the union of a senator and a freed~woman (prohibited by the lex Iulia de
maritandis ordinibus of 18 Be; see Digest 23. 2. 44, Paul; Treggiari, Roman Marriafe,
see Digest 38. II. I. pr. (Ulpian); Treggiari, Roman Marriage, 49-51. The law seems
to have treated the couple as husband and wife, and afforded.them at least same of
the legal rights normally associated with lawful Roman marnage: .e.g. a ~an could
charge his 'wife' with adultery even if they were not lawfully marned (Digest 48. 5·
14. I, Ulpian). .
IIZ Gaius, Inst. I. 78. Examples in Livy 26. 34. 6-9, 38. 36. 5-6 (Campama); 43· 3·
1-4 (Carteia), on which see R. Syme, Colonial Elites: Rome, Spain and the Am~ricas
(London, 1958), 11; M. Mirkovi6, 'Die Entwicklung und Be~eutun~ .der Verleihung
des Conubium' in Eck and Wolff (eds.), Heer und IntegratlOnspolztlk, 168.
113 On the la~, see especially R. Böhm, 'Zur lex Minicia', Zeitschrift der Savigny·
St{ftung für Rechtsgeschichte (Romanistische Abteilung), 84 (19~7), 3.63-71; C.
Castello, 'La data della legge Minicia', in Studi in onore di V :4ranglO~RUlz.~~ples,
1953), iii. 301-17; G. Luraschi, 'Sulla date e sui destinatan dell~ lex ~mlcl~ de
liberis', Studia et Documenta Historiae et Iuris, 42 (1976), 321-7o.1t IS notlced bnefly
in 1. F. Gardner, Women in Roman Law and Society (London, 1986), 138, 143,223;
B. Rawson, 'The Roman Family',in B. Rawson (ed.), The Family in Ancient Rome:
 Husbands and Wives in the Frontier-Zone
war (perhaps in 121 BC), "4 it declared that children born to a
marriage for which the partners did not possess conubium should
take the status of their non-Roman ('inferior') parent. "5 It
affected then only one of the four types of marriage of mixed
citizenship, that of a Roman woman and a foreigner who did
not have conubium, which prior to its enactment yielded Roman
children under the law of nations. Both before the law and after it
the children of a marriage for which the partners possessed con:
ubium took their father's status, and the children of a Roman man
and a foreigner who did not have conubium were born foreigners. A
somewhat different set of rules governed marriages of mixed
citizenship contracted by soldiers. Since soldiers could not lawfully
marry, their children were illegitimate, and therefore took their
mother's status under the law of nations. So the children of soldiers
and Roman warnen were illegitimate Roman citizens; those born to
soldiers and non-Roman warnen were illegitimate non-Romans. 116
It is reasonably certain that the Minician law applied to all
Roman citizens, both those who lived in Italy and those who
were resident in the provinces. Some Roman laws that dealt largely
with matters of contract or property did not apply to Roman
citizens outside Italy: so, for example, the Furian law, passed
perhaps in the first century BC, which regulated the release of
sureties and guarantors. I I7 I cannat imagine, however, that this
could have been the case with the Minician law, if for no other
reason than that Gaius describes it as operating in his own day,
when the number of non-Romans resident in ltaly is likely to have
New Per"pectives (lthaca, NY, 1986),23; Treggiari, Roman Marriage, 45-6. Its terms
are summarized in W W. Buckland, A Textbook 0/ Roman Law fram Augustus 10
Justinian, 3rd edn., rev. by P. Stein (Carnbridge, 19 66 ),99- 100.
114 Cf. Castello, 'La data della legge Minicia'; Luraschi, 'Sulla date e sui desti-
natari'.
115 The terms ofthe law are described in Gaius, Ins!. 1. 78; Tituli Ulpiani 5. 8 . An
ur;.dated deeree of the Senate legitimized certain types of marriage cootracted by
mlstake, e.g. where a Roman man married a noo-Roman woman in the belief that
she was a Roman ~itizen; its terms are described in Gaius, Inst. I. 67, 69-74, 87; see
a!s? Behrends, 'DIe Rechtsregeluogen der Militärdiplome', 120; Cherry, 'Lex Mini-
cm , 256-9.
116 See also K. Kraft, 'Zum Bürgerrecht der Soldatenkinder', Historia, 10 (1961)
120-6; M. Mirkovic, 'Die römische Soldatenehe und der "Soldatenstand''', Zeit~
schrift für Papyrologie und Epigraphik, 40 (1980), 259-71; H. Nesselhauf, 'Das
Bürgerrecht der Soldatenkinder', Historia, 8 (1959), 434-42; H. Wolff, 'Zu den
Bürgerrechtsverleihungeo an Kinder von Auxiliaren und Legionaren' , Chiron, 4
(1974), 470-510. lJ7 Gaius, Inst. 3. 122.
Husbands and Wives in the Frontier-Zone I37
been very smalI. Gaius also applies the law (Inst. r. 79) to Latins
who at one time had their own communities and states and were
classified with foreigners; they are probably the Latins who were
resident in Italy in the period before the Social war. ,,8
Gaius indicates also that a marriage of mixed citizenship was
transformed into a lawful Roman marriage if the non-Roman
partner acquired Roman citizenship. "9 Children born to the
marriage after the non-Roman partner had been given the citizen-
ship were Roman citizens. l2o We are not told whether the enfran-
chised were required to relinquish their former citizenship and the
rights and privileges that are likely to have accompanied it in the
laws of their local communities. The Syrian author Bardesanes,
writing probably early in the third century AD, declared that the
Romans 'abolish the laws of the countries they conquer'.'"' A
number of the provisions of the so-called Gnomon of the [dios
Logos-a summary list of regulations, mostly of a financial nature,
originally drawn up under Augustus, amended and augmented by
later emperors, Egyptian prefects,and the Roman Senate, and
designed to be used in administering the emperor's 'private
account' in Egypt-indicate that the enfranchised in Roman Egypt
were subject to the terms of Roman legislation. m
118 Cf. A. Watson, The Law of Persons in the Later Roman Republic (Oxford,
1967), 27-8. . ..
"9 Gaius: Ins!. I. 67-71, 87; see also I. 90, where, mdulgmg m what mus~ be a
kind of lawyers' game, he describes the rules that governed t~e status ~f. cht1d~en
conceived but not yet born when the non-Roman partner recelved the elhzenshlp.
l~O The Roman citizenship might be acquired in a number of ways other than by
birth to two citizen parents. Those who served as magistrates or decurion.s in towr:s
of Latin status, for example, were awarded the citizenship when they IeH office (11
was given also to their parents, wives, children, and sons' children); see eh. 21 of the
Flavian municipallaw (text in 1. Goozalez, 'The Lex Irn~tana: A New Copy of the
Flavian Municipal Law', JRS 76 (1986),154); see also Gams, Inst. 1. 96; P. Garnsey,
Sodal Status and Legal Privilege in the Roman Empire (Oxford, 1970), 266 ..The
citizenship was given to the ex-slaves of a citizen-owner if they were freed III a
manner sanctioned by law; see 1. P. V. D. Balsdon, Romans and Aliens (London,
1979),86. And the authorities, seemingly as a matter ofpol.icy, periodieally enfran-
chised individuals towns even whole peoples; see e.g. Tacltus, Ann. 6. 37 (Ornos-
pades of Maureta~ia, wh; assisted Tiberius in the Dalmat~an war); ~as~ius Die: 41.
24. I (Gades); Pliny, HN 5. 20 (Rusucurru, in Mauretama Caesanensls); Tacltus,
His!. I. 78 (the Ligones); see also the list in Velleius Patereulus I. 14. 1-8.
121 Book of the Laws of Countries; see MacMullen, Changes, 3 3 . . .
122 BGU 1210; text also in S. Riccobono et al. (eds.), Fantes Iuns Romanz
Anteiustiniani 2nd edn. (Florenee, 1940-3), i. 99. Provisions: e.g. 23, 27-8, 33,
61. The law that governed the intermarriage of Romans and Egyptians is reviewed
 Husbands and Wives in the Frontier-Zone
The status of marriages of mixed citizenship in locallaw is not
attested. '23 Of the Roman jurists, only Gaius shows any interest in
non-Roman law, and then only when it was likely to become an
issue in Roman law, for example when the Roman citizenship was
awarded to one partner of a non-Roman marriage. Cicero c1aimed
(Ba/h. 20) that provincial communities had the right to adopt or
reject Roman legislation, and he was probably correct. It is cer-
tainly hard to imagine that any non-Roman community would
have chosen to adopt the Minician law. The intermarriage of
Roman and non-Roman is likely to have been valid then, or at
least not penalized, in the laws of most non-Roman communities.
In keeping with the temper of much of the Roman law of
personal status, the Minician law reserved its harshest penalties
for the children of the disobedient. Denied the Roman citizenship
and therefore also access to the remedies of the civillaw, they were
probably most aeutely disadvantaged in the Roman law of succes-
sion. They could accept nothing bequeathed to them by their
Roman parent, because non-Romans could not receive inheri-
tanees or legacies from Roman citizens. '"4 Anything left to them
by their Roman parent could be eonfiscated by the state. '25 The
rule extended so far that a Roman could not bequeath even a
usufruct to a non-Roman. 126
It may be doubted, however, that the Minician law was always
rigorously or evenly enforced. It seems to have established no
method of exacting what it prescribed. The status of children
born of marriages of mixed citizenship is unlikely to have become
an issue unless they were betrayed to the authorities or laid claim
to Roman status in a Roman court of law or before a Roman
in Cherry, 'Lex Minicia', 260~2. Examples of Roman-Egyptian marriage are col-
lected in R. Taubenschlag, The Law 0/ Greco-Roman Egypt in the Light of the
Papyri, 2nd edn. (Warsaw, 1955), 106, J08.
1~3 Ulpian does say (Digest 50. I. 1. 2) that children born to the marriage ofthose
who were from different non-Roman communities took their father's status. He goes
on to report that the children of such marriages were sometimes allowed to take
their mother's status; the privilege was granted to the Ilienses, to Delphi, and, by
Pompey, to the Pontici. Some jurists thought that only illegitimate children in
Pontus were eligible; Ulpian and Celsus disagreed. 1~4 Gaius, Inst. 2. 1I0.
1~5 Details in D. Johnson, The Roman Law 0/ Trusts (Oxford, 1988), 35-9.
J~6 Fragmenta Vaticana 47a. Nothing survives to describe the rules that governed
succession to the estate of the non-Roman partner. I suspect that the Iaws of
provincial communities are unlikely to have barred the non-Roman partner from
bequeathing his (or her) estate to his (or her) non-Roman child.
Husbands and Wives in the Frontier-Zone 139
official. There were probably also many local differences in the
degree to which the law was enforced, varying according to the
extent to which Roman law was known and applied. And the
children's disabilities in the Roman law of succession could be
dodged. A Roman could leave property to his non-Roman child
in a trust (fideicommissum), for example, by appointing someone
his heir with the request that the heir pass on all or part of the
estate to the child. '27 Trusts were made actionable by Augustus. It
was later decided, probably by Vespasian, that non-Romans could
not receive trusts from Roman citizens. 128 But other dodges were
at hand, including the secret or tacit trust. "9
The law also allowed a soldier who was a Roman cll1zen (a
legionary, for example) to appoint his illegitimate children his heirs
or legatees. '30 There is no evidence to describe the rules that
governed the succession of illegitimate Roman children to the
estate of a non-Roman saldier. They are, I think, unlikely to
have been disqualified. The same can be said of the illegitimate
non-Roman children of a non-Roman soldier. And soldiers who
were Roman citizens and made a military will were expressly
allowed to make non-Romans their heirs or legatees. '3' The intro-
duction of the privilege cannot be dated. It may be coeval with the
introduction of the military will. Ulpian says (Digest 29. l. l. pr.)
that it was Julius Caesar who first relaxed the rules that governed
soldiers' wills, but adds that the concession was short-lived (con-
cessio temporalis erat). The privileges of the military will were
reintroduced, it seems, by Titus, reaffirmed by Domitian, perhaps
expanded by Nerva, and entrenched in law by Hadrian.'3 2
So there is little reason to think that the law will have deterred
Romans from marrying non-Romans. There is certainly nothing to
indicate that the union of Roman and non-Roman was considered
12
7 Gaius, Inst. 2. 275, 285. On trusts, see D. Daube, Roman Lmv: Linguistic,
Socia! and Philosophical Aspects (Edinburgh, 1969), 96-102; Johnson, Law 0/
Trusts. 128 See Johnson, Law 0/ Trusts, 21-31, 36-9.
1~9 So Jerome, Ep. 52. 6: 'we cheat the law by trusts' (per fideicomf!1issum !ef!ib~s
inludimus). The secret trust is described in Digest 34. 9. JO. pr. (GalUs); detaIls 10
Johnson, Law of Trusts, 42-68. . , .
13° The inheritance rights of illegitimate children are dlscussed at great length 10
the Code o' Justinian 5. 27' summary in 1. Crook, Lmv and Life of Rome (London,
v , 131 • I
1967), 107. . .. Gams, n~t .. 2. JIO.
13~ See V. Arangio-Ruiz, 'L'origine deI testamentum mlhtIs e la sua pOSJZlone nel
diritto rornano classico', Bullettino deI Istituto di Diritto Romano, 18 (1906), 157-96.
140 Husbands and Wives in the Frontier-Zone
to be morally objectionable. Tacitus reports without comment the
mtermarnage of Romans and Germans at Cologne (Hist. 4. 65)
and at Cre~ona (Hist. 3. 34). '33 And the diplomas which the
5
authontles Issued to auxiliary and other veterans awarded them
the right to marry non-Roman women. It may be concluded that Consequences of the
the law IS hkely to have had little, if any, effeet on the incidence of
mtermarriage. Roman Occupation
'33 See also Suetonius, Iul. 52. I, on Julius Caesar's affair with Eunoe 'the Moor',
wife of King Bogudes.

There is little reason, then, to think that the acculturation of the

fron tier-zone in Roman-era Algeria was aecomplished by the inter-
marriage of Roman and north African. There is in fact no evidenee
to show that there was any really significant measure of eultural
change in the region during the period of Roman oecupation. It
might be supposed instead that the main consequenees of the
coming of the Romans were economic and social.
It has been argued of other frontier-provinees that the needs of
the Roman army resulted in the expansion of loeal agricultural
production, and that the army's continuing presence mayaIso have
stimulated the development of long-distanee trade, in so far as
soldiers with cash to spend ereated new and ready markets for all
kinds of goods, at least some of which could not be produeed
locally.' Several lines of inquiry are suggested. Was there a
sustained and measurable increase in cultivation in the frontier-
zone in the period of the Roman occupation? How much of any
increase in production can be attributed to the demands of the
army? How much of it, if any, was accomplished by Roman
colonist-settlers (veterans, for example)? Can we distinguish
increased production that responded to army-demand from what
may have been longer term (that is, indigenous) trends towards
expanded production? Did the whole of the indigenous population
profit from military spending, or were the benefits limited largely

, See e.g. S. K. Drummond and L. H. Nelson, The Western Frontiers of Imperial

Rome (Armonk and London, 1994), 9; cf. H. Elton, Frontiers of the Roman
Empire (BJoomington, Ind., 1996), 8, 83; C. R. Whittaker, 'Rural Labour in Three
Roman Provinces', in P. Garnsey (ed.), Non-Slave Labour in the Greco-Roman
World (Cambridge, J980), 81. The issues are carefully reviewed in Whittaker,
Frontiers of the Roman Empire: A Sodal and Economic Study (BaItimore, 1994),
99- 13I.
 Consequences of the Roman Occupation
to the merchants and camp-followers who supplied goods and
services to the soldiers? To what extent did the Roman occupation
change patterns of landownership or otherwise effect a redistribu-
tion of wealth in the fron tier-zone?
The evidence that might allow us to assess the economic and
social consequences cf the Roman occupation is in same cases
missing, in many others preserved only in fragments, scattered
over space and time, difficult of interpretation.' It hardly needs
saying that I have no perfeet answers. Comparative evidence from
better-documented frontier-provinces, especially Britain and Tri-
3 perhaps Britain). It isuneertain, too, how much of tbe food and
politania, may sometimes be used to define the limits of what the
Romans can be expected to have accomplished (or to have
destroyed) in Algeria. I am certain, however, that the picture
described here wilI need to be modified when there is systematic
investigation of rural sites in the frontier-zone of the sort for
example, that has been carried out in Tripolitani; by the UNESCO
Libyan Valleys Survey project4
AGRICULTURE
It has been said recently of the western frontier-provinces that the
army's 'steady appetite' for food and other materials was 'the
fundamental factor that led to the development of the frontier dis-
tricts'; the indigenous population, we are told, was 'forced to pro-
duce a surplus', most ofwhich 'went directly to supplying the army'. 5
A similar argument has been advanced by Ramsay MacMullen: 'the
.2See also B. D. Shaw, 'The Structure of Local Society in the Early Maghrib: The
Blders', The Maghrib Review, 16 (1991), 19.
3 Recent work 011 the [rontier-zone in Britain is described in D. 1. Breeze, 'The
Nortllern Frontiers', in M. Todd (ed.), Research on Roman Britain [9 60- 89 (Landan,
8 Classical Antiquity (Cambridge, 1983), 124-5 n. 4I.
19 9), 37-60. On Tripolitania, see especially D. 1. Mattingly, Tripolitania (Ann
Arbor, 1994).
4 Th: pr~ject is described in G. D. B. Jaues and G. W W Barker, 'Libyan Valleys
Survey, üb. Stud. 11 (1979-80), II-36; Jones, 'The UNESCO Libyan Valleys
Survey; The Development of Settlement Survey', in D. 1. Buck and D. 1. Mattingly
(eds.), Town and Country in Roman Tripolitania: Papers in Honour o/Olwen Hackett
(Oxford,. 19 85), 263-89;. see also Barker, 'The UNESCO Libyan Valleys Survey:
Developl~g MethodoJogles for Investigating Ancient Floodwater Farming', in Buck
a~d Mattmgly (eds.), Town and Counlry, 291-306; Barker and Jones, 'The UNESCO
Llbyan Valleys Survey UI, Palaeo-Economy and Environmental Archaeology in the
Pre-Desert', Lib. Stud. 13 (1982), J-34.
5 Drummond and Nelson, Western Frontiers, 9. See also C. M. WeHs, 'The Impact
Consequences of the Roman Occupation 143
demands of resident troops for agrieultural products of all sorts
might enlarge and nourish the local economy, through requiring
more land to be farmed and surpluses to be produeed'. 6
There ean be little doubt that the army as a whole consumed a
great deal of grain and other agrieultural produets; the European
frontier-armies alone are thought to have requrred the skms of
three-quarters of a million calves to make and to repair their
tents. 7 It is not at all clear, however, that army-demand had a
significant impact on the economy of any given region (except
other materials required by the army was supplred by loeal
farmers and how mueh was produeed by the soldiers themselves.
Widely distributed inseriptions that mention military conductores
('eontractors') and pecuarii ('cattlemen') indieate that required
supplies were sometimes obtained from independent clvllran
producers 8 The ostraka reeovered at Bu Njem suggest that it
was eustomary for the soldiers posted there to obtam food from
loeal agriculturalists 9 At least some food was grown on the
military land (territorium or prata legionis) that extended, m
Roy Davies's words, 'a considerable distance' around most
of the Augustan Campaigns on Germany', in D. M. Pippidi (ed.), Assimilation:1
resistance ci la culture greco-romaine dans le monde ancien: Travaux de VIe. Congres
International d'Etudes Classiques (Madrid, Sept. 1974) (Bucharest and Pans, 1976),
422. There is an excellent discussion of the army's needs in D. 1. Breeze, 'Demand
and Supply on the Northern Frontier', in R. Miket and C. Burgess (eds.), s.et,?e~n
and beyond the Walls: Essays on the Prehist01Y and History 0/ North Entam m
Honour 0/ George lobey (Edinburgh, 1984), 268-72.
6 Changes in the Roman Empire: Essays in the Ordinary (Princeton, 1990), 58:9;
see also W. Groenman-van Waateringe, 'Urbanization and the North-West Front~er
ofthe Roman Empire', in W. S. Hanson and L. 1. F. Keppie (eds.), Roma~ Frontter
Studies 1979 (Oxford, 1980), 1037-44; C. R. W~ittaker, 'Trade and FrontIers ~ft~e
Roman Empire', in P. Garnsey and C. R. WhIttaker (eds.), Trade and Famme In
.
7 Drummond and Nelson, Western Frontiers, 80. See also C. ~. Whlt~ak~r,
'Introduction', in C. R. Whittaker (ed.), Pastoral Economies in Classlcal AntlqUlty
(Cambridge, 1988),4, on cattle products. . .
8 C. R. Whittaker, 'Agri Deserti', in M. 1. Finley (ed.), Studles In Roman Property
(Cambridge, 1976), I58-9.
9 R. Marichal, Les Ostraca de Eu Njem (Tripoli, 1992), 99-106; see also. ~at­
tingly, Tripolitania, I52. Soldiers with apermit (diplom,a) coul~ lega.lly reqUIsltlOn
supplies from civilians, usually, it seems, a! a set pnce, ~hlCh mIght be mor:,
normally not less, than the current market-pnce; see Breeze, Demand and Supply,
277' R. W Davies, Service in the Roman Army, ed. D. Breeze and V A. Maxfield
(Ne~ York, 1989), 51; R. MacMuHen, Soldie~ and Civilian. in the Later Roman
Empire (Cambridge, Mass., 1963), 85-9; cf. Wluttaker, FrontIers, 106.
 '44 Consequences oj the Roman Occupation
camps. <0. The land seems sometimes to have been worked by
the soldlers themselves-we may recall the legionaries harvest-
mg hay (or overseeing its collection) at Casae-but was prob-
ably more often leased to. civilians." It is unlikely. however.
that large leglOnary bases (hke Lambaesis) could have obtained
the food and fodder that they required entirely from military
land.
The wealth of north Africa in the Roman period. as in the pre-
Roman era, was almost entirely in agricultural products (not in
industry, which was always small-scale)." It may safely be said
then that any significant measure of economic growth could have
been achieved only by expanding agricultural production. There
seems now to be little doubt that production was increased, perhaps
dram~tlcally, acros~ much of north Africa; the Roman accomplish-
ment, It has been sald, was 'vastly to extend the area of cultivation'. 13
Production was expanded in the Aigerian fron tier-zone from at
least the early second century AD, mainly, it appears, in the steppe-
lands that lay behmd (that IS, to the north 01) the frontier-line. '4 It
might be conjectured therefore that increased production was a
consequence of the creation of the fron tier, and of its linear bar-
riers (thejossatum), which, in Eric Birley's opinion, were designed
(m part) to mark the southem limit of economic development. '5 It
has often been remarked that the density of farm-settlements in
)0 Da~ie~: Service, 188. See also A. M6csy, 'Das territorium legianis und die canabae in
Pa~nOll1en, Acta Archaeolog;ca :1-,cademiae Scientiarum Hungaricae, 3 (1953), J79-99.
Casae: eIL 8. 4322. On mIlItary land leased to civilians, see Breeze 'Demand
al1d Supply', 27?; cf. N. 1. Higham, 'Roman and Native in England N~rth of the
Tees: AcculturatlOll and !ts Limitations', in 1. C. Barrett, A. P. Fitzpatrick, and L.
Macm~es (eds.), Barbanans and Romans in North-West Europe: Frorn the Later
Rer;.ubilc to Late Antiquity (Oxford, 1989), 161.
13 Cf. S. Raven',Rome in Africa, 3rd edn. (London and New York, 1993),9 .
6
C. M. WeHs, The Problems of Desert Frontiers', in V. A. Maxfield and M. 1.
Dobson (eds.), Roman Frontier Studies 1989: Proceedings ofthe XVth International
~ongress of R~man Frontier Studies (Exeter, 1991), 479. See also D. J. Mattingly,
New Per~peC!lve~ ?n the Agricultural Development of Gebel and Pre-Desert in
.R0m~n ~np~htama, ROMM 41-2 (1986), 45; Tripolitania, 138 (a 'massive increase'
In T~lpohta~la);, D. 1. Mattingly and R. B. Hitchner, 'Roman Africa: An Archae-
ologlCal RevIew, JRS 85 (1995), 204.
i4 C. Daniels, 'The Frontiers: Africa', in 1. S. Wacher (ed.), The Roman World
(L?5nd~m an~ Ne~ Y<;,rk, 1987), i. 235; Raven, Rome in Africa, 91.
BI:ley: Hadnamc FrontIer Policy', in E. Swoboda et al. (eds.), Carnunlina:
Er~eb~lsse de~ Forschung über die Grenzprovinzen des römischen Reiches: Vorträge
beim mternatIOnalen Kongress der Altertums/orscher Carnuntum 1955 (Graz and
Cologne, 1956), 29-32.
Consequences oj the Roman Occupation '45
Roman Britain was much higher in the region south of Hadrian's
wall than in the area that lay beyond it. ,6 The problem is that
virtually nothing is known of agricultural settlement in the period
before the building ofthe wall (or, for that matter, ofthejossatum),
so that variations in settlement-density in the Roman period can-
not be shown to be a consequence of its construction (or of the
Roman presence). 17
There is no evidence to indicate that there was any real change
in the kinds of crops that were grown in the Aigerian fron tier-
zone. Grain production may have been increased, perhaps at the
expense of pasture. Pastoralism was perhaps also intensified, with
the result that in some areas at least it may have developed into
long-range nomadism. ,8 But the principal agent of agricultural
expansion was the olive. A subsistence crop, it seems, in the pre-
Roman era, its widespread cultivation in the frontier-zone in t~e
Roman period, and perhaps especially in the third century AD, IS
attested by the recovery of numerous ruins of olive-presses in the
valleys of the Aures mountains and in the plain-Iand to the north
of them. '9
It is reasonably clear, then, that there was an increase in
agricultural production in the fron tier-zone in the period of the
Roman occupation. The real question is whether any of the
increase can be attributed to the coming of the Romans. There
are, it seems to me, three ways in which the Romans might have
contributed to the expansion of cultivation: by introducing new or
better methods of farming and/or of water-management; by
increasing the demand for agricultural products; or by promoting
the cultivation of marginal land. For the first, at least, there is no
evidence at all.
The notion that the Romans introduced better techniques of
16 See e.g. G. D. B. Jones, '''Becoming Different Without .Knowing It":. ~he Role
and Development of Vici', in T. F. C. Blagg and A. C. Kmg (eds.), Milüary and
Civilian in Roman Britain: Cultural Relationships in a Frontier Province (Oxford,
1984), 86; Whittaker, Fronliers, 86-7. . . , .
17 D. 1. Breeze, 'The Impact of the Roman Army on North Bntam, III Barrett,
Fitzpatrick, and Macinnes (eds.), Barbarians and Romans, 231.
18 E. W. B. Fentress, 'Forever Berber?', Opus, 2/I (I983), 164, 167.
19 See ibid. 165. On the importance of the olive, see also H. Camps~Farber,
L'Olivier el l'huile dans I'Afrique romaine (Algiers, 1953); D. 1. Mattingly, 'First
Fruit? The Olive in the Roman World', in G. Shipley and 1. Salmon (eds.), Human
Landscapes in Classical Antiquity: Environment and Culture (London and New
York, 1996), 213~53.
 '4 6 Consequences 0/ the Roman Occupation
Consequences of the Roman Occupation 147
cuItivation and of water-retention and distribution in north Africa .' . 26
re indigenous. The same can
has been refuted (by Brent Shaw and others). But it refuses to go Thelepte in Roman-era TumsIa we 10 ed in the pre-desert
away altogether. So it has been argued recently that the agricul- be said of the cultivation techmques emp Y allel perhaps,
T' 1"t . 27 There IS an mterestmg par ,
tural 'boom' in north Africa in the Roman period cannot be zone of npo lama. , dh the 'desert' or 'thirsty' people of
explained 'if agrarian technologies or systems of water use were among the Tohono 0 0 arr:;;ern Sonora (Mexico), who, long
not changed,.20 Even C. R. Whittaker, who knows better than southern Anzona and no E ans cultivated maize, beans,
h ·ng of the urope , . c
most historians the biases that inform a great deal of colonialist- before t e cornI d ourds in a region notonous lor
era scholarship on Roman north Africa, says only that 'many of squash, cotton, tobacco, a~ g th f the wash' (akchin), where
the supposed agrarian innovations in Africa, being undated, could its aridity, by farmlllg the f;nou 0 as carried over the land; in
be pre-Roman.''' It is sometimes suggested, too, that the agricul- arroyos widened and run-o water w 11 and surface-water were
t rritories where ram fa ff
tural development of the fron tier-zone was accomplished by veter- their eastern e , '1 d' . n walls to capture run-o ,
b dant they bm t IverslO - . ,8
ans who settled in the region after they were discharged, bringing more a u n , d b h structures to reduce erOSlOn.
wirh them 'more intensive farms of production'. 22 So the extension and low earthen dykes a~ r r~sei~her tbat Roman veterans were
of agriculture in the Auros mountains in the second century AD, in There IS no reason to e JeV t f the fron tier-zone. It is hardly
the valleys of the Oueds el-Abiod, el-Abdi, and Rassine, is said to responsible for the developmen 0. ved techniques of cultiva-
have been a product of veteran settlement. 23 likely that veterans learned new .~: I~::: has pointed out that the
It was remarked earlier that many of the rural water-control tion dunng theIr penod o;.~e~v\o· have discharged annually more
schemes of Roman-era Algeria must necessarily have been legion at LambaesIs IS un 1 e Yb be expected to have settled
ly some ofw om can . f
constructed before the coming of the Romans because the than 100 men, . h e concI u des, an 'extreme pauclty 0
f . onneo there IS
centuriation-lines that are visible on aerial photographs were in the rontler-zo, , f ' 29 And even where veterans
drawn across the hydraulic systems. 24 Much of the fossatum, evidence for supposed vetera: armers; be shown to have had any
for example, was built on top of irrigated field-systems: in the are known to have settled, t ey canno 30 The expansion of agri-
Jebel Mekriziani-Seba M'gata region; near the encampment that discernible effect on the local economY;h Africa is very likely to
cultural production in Roman-era nor
Jean Baradez called Fort Parallelogramme; and in the Hodna
mountains. And at least some parts of the Roman-era water-
control systems lay outside the zone of Roman political controI. 25
Bruce Hitchner has demonstrated that the dry-farming methods
. 1 Surve 1982-1986', Ant. afr. 24
practised in the rich agricultural area around Cillium and 26 Hitchner: 'The Kasse.rine Archae~lo?~ct Survey y'1987', Ant. afr. 26 (r99 ),
(1988), 7-41; 'Thc Kassenne Archaeo oglca , 0

1
23 -59. , d' t'Limesetmaitrisedel'eau',ROMN!1 1-
20 A. Carandini, 'Pottery and the African Economy', in P Garnsey, K. Hopkins, 27 P. Trousset, 'De la mon.ta~ne au e(~~here Archaeology and Military Trammg
and C. R. Whittaker (eds.), Trade in the Ancient Economy (Berkeley and Los 2 (19 ), 95· R. G. Goodch1~? s theoduard" Outposts in Tripolitania', Report.\·, and
86
Angeles, 19 83), 157; cf. R. I. Lawless, 'L'Evolution du peuplement, de I'habitat et Go Hand in Hand: Roman Home . .. 'n Tri olitania 2 (1949), 29-35; The
des paysages agraires du Maghreb', Annales de geographie, 81 (197 2), 4 0.
6 Monographs of the Department of ~ntl~~l;l:~~t the ~xpansidn of cultivation in the
21 'Supplying the System: Frontiers and Beyand', in Barrett, Fitzpatrick, and
0
Limes Tripolitanus', JRS 4° (1950 , 3 IOIII·sts is denied in 1. R. Bums and B.
Macinnes (eds.), Barbarians and Romans, 68 (my italics); cf. 'Land and Labour in r
pre-desert was accomp lS . Dhed by Roman ' .coThe Tripolitanian Example, 3?0 BC-i\D
North Africa', Klio, 60 (1978), 331. Denness, 'Climate and Soclal (Ydna)m1~s. alld Country 217- 1 9; Damels, 'The
22 E. W B. Fentress, Numidia and the Roman Army: Social, Military and Eco-
,. k d M ttingly e s. , .J OIvn ,
3°0, m Buc an. a . 1 'New Perspectives', 47.
nomic Aspects 0/ the Frontier Zone (Oxford, 1979), 177. Frontiers: Afric~', 1. 2~0-2; ~Tatlftll~ y: Dog' American Indians, Environment, and
28 D R LeWIs Nezther rrO 6
23 T. R. S. Broughton, The Romanization 0/ A/rica Proconsularis (Baltimore,
01.
19 29), 140; cf 128.
Agrari~n Change (New York .and ~~fot;~~~~\..~·~I~i~i~' Opus, 211 (1983), I40.
24 B. D. Shaw, 'Water and Society in the Ancient Maghrib: Technology, Property 29 Shaw: 'Soldiers and, Soclety: . e Ancient Irrigatio;l Community',. A~t. afr.
and Development', An!. afr. 20 (1984), 12 7. 3° Ibid 140' see also Lamasba. An
25 Ibid. 157, 162. 82 . , . . " H D'Escurac D. OlSy, 'L e Phenomene assoclatlf dans
r8 (19 ), 8811. 2,. 9?: pace
le monde paysan a 1 epoque u ci. Haut-Empire', Ant. afr. I (19 67), 61.
 148 Consequences of the Roman Occupation
have been, as Rober! Broughton once remarked, 'primarily a
natIve development'. 31
The only question then is whether the increase in production
was a consequence of Roman demand for agrieultural products,
eüher for the soldIers statlOned in the fron tier-zone or for military
and clVIhan markets in other parts of north Affica or in Europe. It
ought to be smd at the .outset that, in the present state of the
eVIdence, no enlIrely sallsfactory answer can be formulated. It
does seem to me, however, that there are some reasons for thinking
that the effect at least of frontier-zone army-demand may have
been smaller than has sometimes been suggested.3'
It. se.ems now to be generally agreed that agriculture in the
empIre s f~ontIer-zo~es was 'stimulated' by the army's consider-
able appellte for gr~m and fodder. 33 Whittaker, for example, has
suggested. that the declslve factor' in economic development on
the fronlIers was 'the fron tier itself and the large consumer
market m the form of the Roman army' 34 It is, however, almost
ImpOSSIble to demonstrate that agricultural production in any
part of the fronlIer-zo~e was mtended, exclusively or even pri-
manly, to feed the soldIers statlOned in the region (and/or their
alllmals). And there is no way to measure the effect of military
demand.
Elizabeth Fentress once tried to calculate the grain requirements
of the legIOn at Lambaesis and their consequences for agricultural
productIOn m the regIOn 35 Assuming that the 5,000 or so legion-
3
1
Broughton: Romanization o"AI'rica 225 .
32 ~ J ,
34 ~.g. ,by Fentre~.s, Numidia, I2?, 33 Raven, Rome in Africa, 77 .
. Rural LaboUl: ,?I. Cf" M. ~enabou, La Resistance africaine Ci la romanisation
(Pans, 197 6), ~9:, llOtenslficatlOll de l'activite economique est fonction de la
dema~de romame; Fe~tress, Numidia, 176: the army was a 'powerfu! stimulant
to agllcultural productlOn'; M. 1anon, 'Remarques sur la fron tiere de Numidie' in
M,axfield and Dobson (eds), Rom~n ~ron.tier Studies /989, 483: it is 'very possible'
~hat the development of ohve cultlvat1.on In th~ region around Mena'a was directly
1elate~ ~~,the pr:s~nce ofthe fort ~nd 1tS assocmted settlement (vicus); Y. Le Bohec,
~~ !f~teme LegIOn Auguste (Pans, ~989), 578: legionary fortresses functioned as
venta e~ cen~res: : .. de conso,mmatlO?'? Matting.ly, Tripolitania, 140 ; the Roman
prese-?ce In Tnpohtal1la was a catalyst 111 extendl11g agriculture south of the 150
mm., lsoh~et ~ee also N, ROYl~ans, 'The North Belgic Tribes in the Ist Century
B.e.. A Hlston~al-~nthropologlcal Approach', in R. Brandt and J. Slofstra (eds.),
Roman and NatIve In the Low Countries: Spheres o/Interaction (Oxford 19 83) 8
who goes a step further in .sugge~ting that the army on the Rhine, becaus~ of its ~;ed
f?~ f~od an,d ~ther matenals, will have tried 'to stimulate tribaI economic produc-
tlVity (my ItaiJcs). 35 H 'd"
l~uml Ta, 125.
Consequences of the Roman Occupation 149
6
aries will each have eaten a kilogram of wheat per day,3 she
estimated that the legion's ammal grain consumption was 1, 82 5
tonnes. Borrowing Jean Despois's figures for modern sowing-rates
in the region (50 kg./hectare) and for expected average crop-yields
(1:5),37 and allowing for grain held over each year for seed, she
suggested that in an 'average' year 91.2 square blometres would
· reqmre
have been needed to produce the grain that t h e IegIOn . d .3 8
Assuming further that one household could farm up to 20 hec-
tares and that it would need about 6 hectares to grow food for Ils
own 'eonsumption, she calculated that the legion could have been
fed by the 'surplus' production of about 650 households. Her
conclusion according to which a peasant household fed almost
,
eight legionaries, . an ,economlC
is, as Shaw has put It, . ab surd't'
1 y .
39
But even ifher other figures were eorrect (or nearly SO), they could
not really be said to improve our understanding of the practieal
effects of army-demand. It would surely be wrong to suppose that
90 - odd square kilometres of previously uneultivated land were
brought into production to feed the legion. Some of the reqmred
increase must have been accomplished by more intensively exploit-
ing land that was already under cultivation. But how much? And
how are we to determine whether the necessary surplus was drawn
from a large number of grain-farmers over a wide area or, at the
other extreme from a few large-scale producers? There is, in other
words, no wa~ even to estimate the per capita effect of increased
demand in the fron tier-zone, which is surely what really mattered
on the ground. 4 0
I am not suggesting that the presence of the Roman army had no
effect on agricultural production in the fron tier-zone. My po mt
36 The fiaure corresponds to about 3 Roman pounds ofbread (I Roman pound =
roughly 3;7 g.; see R. P. Duncan-Jones, The Economy 0/ tI:e ~oman Emp~'re:
Quantitative Studies, 2lld edn. (Cambridge, 1982), I47 o. 2), WhlCh 15 what soldlers
were issued in the 3rd ceot. AD (Fentress, Numidia, I45 o. 2). Cf. Br~eze, 'Demand
and Supply', 269, who estimates that a typical soldier ate about a thlrd of a ton of
grain annually; Elton, Frontiers, 67 (1.5 kg/day). .
37 Despois: La Tunisie orientale, Sahel et basse steppe, 2nd edn. (Pans, 1955),23 8.
3& M. Millett The Romanization 0/ Britain: An Essay in Archaeological Interpre-
tation (Cambridge, 1990), 57, conc1udes that the army in Britain could ha~e. been
fed by the tax product of an area of 28-45 ~m2; cf. Mille.t~, 'Forts an~ ~l~e Ongms of
Towns: Cause or Effect?', in Blagg and Kmg (eds.), Mllltary und ClVlhan, 68-7 2 .
39 Shaw' 'Soldiers and Society', I56 n. 48.
40 Cf.C: R. Whittaker, 'Da Theories of the Ancient City MatterT, in T. 1. Cornell
and H. K. Lamas (eds.), Urban Society in Roman !taly (Landon, 1993), 3·
 r 50 Consequences 0/ the Roman Occupation
rather is that there is no way to isolate production for the army in
the archaeological record, and therefore no real basis for measur-
ing its impacl. No one, I think, would argue that the army's
presence is in itself enough to explain the dramatic increase in
production that is attested over much of north Africa 4 ' Apart
of the increase must have been intended for external markets (at
least some of which may have been opened up as a consequence of
the Roman conquest).4' How are we to weigh their effect on
production against the impact of military demand or even against
what may have been longer-term trends toward increased cultiva-
tion? The more intensive system of production that is attested in
the region of the Rhine in the first two centuries AD, for example, is
now thought to be at least partly a consequence of increased
demand in the indigenous population. 43 The extension and inten-
sification of agriculture in Britain seems to have occurred over a
period much longer than that of the Roman occupation. 44 Where
pollen diagrams used to be taken to indicate an increase .in pro-
dnction in the highlands in the Roman period, new diagrams and
the redating of some old ones show that the increase in cultivation
probably belongs mainly to the (pre-Roman) Iron Age and to the
later years of the Roman presence. 45
There is also little reason to believe that the imperial govern-
ment actively encouraged agricultural expansion in north Afriea.
1
4 cr. Mattingly, 'New Perspectives', 60, on the Tripolitanian pre-desert in the Ist
and 2nd cents. AO ('a few forts can hardly explain the full picture of agricultural
development').
2
4 Cf. G. D. B. Jones, 'Concluding Remarks', in Buck and Mattingly (eds.), Town
and Country, 310, who argues that the expansion of agriculture in Tripoiitania in the
Roman period is 'best seen in terms of an indigenous response to the growth of
external market farces'; see also G. W. W. Barker and G. D. B. lones, 'The UNESCO
Libyan Valleys Survey VI: Investigations of a Romano-Libyan Farm', Lib. Stud. I5
([9 84), [-44·
43 L. Hedeager, 'Empire, Frontier and the Barbarian Hinterland: Rome and
Northern Europe from AD J-400', in M. Rowlands, M. Larsen, and K. Kristiansen
(eds.), Centre and Periphery in the Ancient World (Cambridge, J987), 134.
44 M. lones, 'Crop Production in Roman Britain', in D. Miles (ed.), The Romano-
British Countryside: Studies in Rural Settlement and Economy (Oxford, 1982), i. 99;
see also lones, 'Agriculture in Roman Britain: The Dynamies of Change', in Todd
(ed.), Research on Roman Britain, 134 ethe Roman occupation took place in the
context of mueh lünger-term change in the agrarian landscape'); D. Miles, 'The
Romanü-British Countryside', in Todd (ed.), Research on Roman Britain, 124 (on
the Fenlands); V. A. Maxfield, 'Conquest and Aftermath', in Todd (ed.), Research
on Roman Britain, 29.
45 C. S. Sommer, The Military Vici in Roman Britain (Oxford, 1984), 39.
Consequences 0/ the Roman Occupation I 5I
A Hadrianic law, which appears to have restated the provisions of
a lex Manciana of unknown date, was eVldently llltended to pro-
6
mote the cultivation ofmarginalland (subseciva)4 But it probably
applied only to imperial estates, most of which, it seems, were III
the rich Bagradas valley south-west of Carthage and III the regIOn
around Sitifis 47 The law made various arrangements about share-
cropping,4 8 awarded the right of pos session (ius possidendi) to
those who undertook to work marginal land that had gone out
of cultivation, and offered them reduced rent on land that they
plan ted with fruit or olive trees, figs, or vines. It used to be thougbt
that the law's purpose was to expand tbe dass of indigenous
peasant-farmers; it is now understood that few amo~g the landless
would have had enougb capital to pay for the plantlllg of Vllles or
trees 49 Tbe law's practieal effect, it may be supposed, was to
transfer possession of some marginal land to well-capitalized, IDeal
landowners. . .
ltis sometimes supposed tbat the army's effect on provlllcIaI
economies can be measured by calculating (or eshmatlllg) Ils
impact on provincial population levels; we are told, for example,
that because the soldiers in Britain were only 2-5 per cent of tbe
provincial population, their presenee 'required an equally small
increase in overall production' .5° By the same reasomng, the
Roman oceupation of north Africa can be expected to have pro-
dueed an increase in agricultural production of perhaps 0.3-0 -4
per cent (assuming 20,000-25,000 soldiers and a regional popula-
tion of 6-8 million). In fact, we mlght suppose that the effect was
4 6 See D. 1. Crawfürd, 'Imperial Estates', in Finley (ed.), Studies i~ Roman
Property, 54 (cf. 49). The Hadrianic law is quoted on an altar found at Am .Oussel
(eIL 8. 264I6; AD 2°9-12); for the lex Manciana, see CI~ 8. 25902 (Hr. MettIch; AD
I l 61r 7); for the dates, P. MacKendrick, The North Afncan Stones Speak (Chapel
Hill, NC, 19 80), 52. . ..,. G . dC
47 P. Garnsey, 'Rome's African EmpIre u~der the Prmclpate .' m P. arnseyan .:
R. Whittaker (eds.), lmperialism in the Anelenl Warld (Cam~ndge, 1978), 233· Pace
Fentress, Numidia, 135, 137-9; 'Forever Berber?:, 167, the~e lS no ~easo~ to suppos~
that a large part ofthe region north ofthe Aur~s mo.untams was Impenal pr?perty,
see Shaw, 'Soldiers and Society', 140. On the Impenal estates of nort~ Afnca, s~e
.. 11 D P Kehoe The Eeonomics 0/ Agrieullure on Roman Impenal Estates In
especmy . . , 8 ' k 'R ILb '8
North Afriea (Göttingen, 1988). 4 See ~hltta er,. ,ura a our, 2.
49 Carandini, 'Pottery', 156; Garnsey, 'Rome's Afncan ~mplre, 234.. .
5° M. van der Veen, 'Native Communities in the Frontler Zon~: Umformlty 01'
Diversity?', in Maxfield and Dobs~n (eds.), R01:lQn.Fron~ier.St.udles 1989, 446. A
similar argument is advaneed in MllIett, RomamzallOn oi Bntaln, 57·
 154 Consequences of the Roman Occupation Consequences 0/ the Roman Occupation 155
~~~~~:~ts.,;,;as in circulation in the period before the coming of SOCIETY

There were no doubt many . h' '

might profit from military sp::~:n; ~t ~~s t~e 10cal.~oP~lation
Arehaeologists and historians of the western provinees who have
attempted in reeent years to assess the effects of Roman eonquest
wealth of Belgie Gaul was a eonse uenee of 't een sm t. at the
produetive opportunities opened ~p by the ~ commereml and
and oeeupation have tended to concentrate on trade and exchange;
presenee of th I' on quest and the far less attention has been paid to the impact of military rule on
. '1 e army a ong Ils frontier in the Rhineland' 58 A indigenous social structures and institutions-and not without
:~~,~r sort of development is likely to have oeeurred in Ro;"an- good reason. It is extraordinarily diffieult to use the artifact reeord
fewer 1:~;~~~r~h~~~i~::~bo~lih~:~r::~~~~caSCnalfe,
sitnee tThhere were
to reeonstruet, say, the relationship between the army and the loeal
population, or to trace changes in indigenous society. Mainly
somet' 1 fon leT. e army
Imes et out contraets for uniforms and other kinds f . beeause of the inadequaeies of the physical (and written) record,
ment. 59 And the civilian settlements that grew up aro 0d eqUlp- a number of models have been developed to deseribe the ways in
~~~~:~:; ~:rt!~~~~P;i~~~a~~ ~~~ep~~1
I . . I
to have provided aU:ri::;r::r
-gammg, taverns brothels 60
which the Roman presenee might have affeeted indigenous soci-
eties. It is now generally agreed, for example, that the Roman
lmagme t lat the opportunities opened up by military ~pe d' '. system of administering the provinces, which aimed at eo-opting
t j le army-towns served over th I n mg m local elites, is likely, over the long term, to have reinforced their
from the countryside6,' e ong term, to draw population
power63 But it mayaiso be conjeetured that the importation of
It i8 impossible. now to determine what rneasure of economic Roman or Roman-style goods in the period after eonquest will
f:~~:~ ~~~;or ~Ight hfiave been) achieved. It is difficult even to have undermined the position of loeal elites by devaluing com-
" e w 0 pro ted most. We might suppose th h modities (wine, for example, in the case of Britain) which, beeause
pnnclpal beneficiaries were the li'aders and m h' hat t e they had previously been searee, had served to confer prestige on
up shop in c .. ere ants w 0 set
frontie ommumlies along the provision-routes and in the the few who could obtain them 64 And wherever there was a
r-zone settlements that mark t d d prolonged period of direct military rule, as in the north Afriean
the camps.62 e e goo sand services to
frontier-zone, tbe army is likely further to have eroded the position
of local elites by usurping at least some of their traditional poli-
57 cr. Elton, Frontiers 82' MacMullen eh
the main source of (coin~d) 'mon . th' f an.ges, 59· The anny was undoubtedly tieal and military functions, and by providing young, indigenous
Fentress, The Berbers (Oxford I ey;n ~ ronher-zones:, s~e M. Brett and E. W. B. males with the opportunity to acquire prestige and influence
Unites auxiliaires de l'armee :m~~iJ~ ~5, ~e~tresspNumldw, ?7S; Y Le Bohec, Les
Haut-Empire (Paris I9 89) 18S' Sh n 'Slfn~ue roconsulmre el Numidie sous le through military service, and thereby to bypass already estab-
, , , aw, oldlers and Soc' t ' Th' lished avenues of social mobility6 5 It may be supposed, then,
reason to suppose, however, with Drum d d le y, 149· ere IS no
tha~ the several increases in soldiers' amonh' an Nelson, Western Frontiers, 167, tbat the Roman oecupation of Algeria (and perbaps that of other
dunng the Ist and 2nd cents p Y w. ICh are known to have been awarded
fron tier'; their real purpose ,', ':SAD] were deslgbned to 'keep fron tier money on the provinces) resulted in a restrueturing of indigenous social hierar-
' ' c ear, was to uy the ar ' I ]
chies; while some north Afrieans undoubtedly profited from the
fi g~re IS taken from Fentress, 'Forever Berber?' 4 my s oya ty. 10,000,000; the
16
C. Haselgrove, 'Culture Process on h P.'
during the Late Republic and E ] E . t, ~ enphery; Belgic Gaul and Rome
Roman presenee by adopting (or mimieking) Roman habits and by
( eds. ) , Centre and Perinhery 108
ar y mpne 10 Rowlands L
'
d' .
, arsen, an Knstransen conforming to Roman expectations, others who would or could
6cCf ' r"
. p, Mlddleton, 'The Roman Arm 59 Wh'" k
, I a er,
F: .
rontlers, J08,
and Whittaker (eds) Trade aIId" . 8yand Long-Dlstance Trade', in Garnsey
61 ' . , , ramme, 2 n. I.
Cf. 1. Mlkl Curk, 'Natives Romans and N .
the 2nd Century (The Role oi th A . ewc.omers 10 the Eastern Alps during 63 See A. P. Fitzpatrick, 'The Uses of Roman Imperialism by the Celtic Barbar-

Dobsol1 (eds.), Roman Frontier St:di;r;I9~n Ethmc Interacti~)ll)', in Maxfield and ians in the Later Republic', in Barrett, Fitzpatrick, and Macinnes (eds.), Barbarians
62 Cf. Whittaker, Frontiers 112 0 9'h250 , on the regIOn around Poetovio. and Romans, 31. 64 Millett, Romanization of Britain, 58.
227, ' . n illere ants, see also MacMullen, Changes, 65 See also Janon, 'Remarques sm la [rontiere', 482; Millett, Romanization of
Britain, 59-60.
 Consequences of the Roman Occupation
not adapt themselves to Roman ways were probably politically and
socially marginalized. 66
The effect of the Roman occupation of the N etherlands, it has
been suggested, was the 'detribalization' or 'peasantization' of
indigenous society; tribai structures, we are told, gave way to a
'peasant society' as traditional bonds, like kinship, 'lost impor-
tance' in the process of integrating native society into the Roman
'state-system,6 7 It is possible, I suppose, that something similar
happened on the north African fron tier. But there is no real reason
to believe that it did. What is known of north African tribai
customs and institutions in the periods before and after the Roman
occupation suggests that there was little real change in the period
of Roman rule. 68
lt is regrettable that the views and practices of indigenous
women in Roman north Africa (as in most parts of the Roman
world) are inaccessible. Largely undocumented in the written
sourees, their presence in the archaeological record normally indis-
tinguisbable from that ofmen, they are only occasionally attested-
as names-on inseriptions. MacMullen has suggested that provin-
cial women 'on the road to Romanization' are likely to have
travelled 'at a different pace from their men-generally more
slowly'. They were, he goes on to say, probably less quick to
abandon traditional forms of dress and of speech, at least in part
because they travelled less than men, appeared less often in public,
and generally had fewer opportunities to talk with strangers. 69
Some measure of corroboration is provided perhaps by Cicero,
who claimed (De Orat. 3. 45), on what basis it is difficult to say,
that 'it is easier for wornen to preserve old ways intact, because,
knowing little ofthe different kinds of speech, they hang on forever
to what they first Jearned' Uacilius enim mulieres incorruptam
Consequences of the Roman Occupaaon 157
antiquitatem conservant, quod, multorum sermonis expertes, ea
tenen! sempe1: quae prima didicerunt). I am not sure how ~uch
importance we ought to attach to the fact that on north Afncan
inseriptions a preponderance of those altested with indigenous
narnes are wornen. 70
I suspect that the Roman occupation of north Africa may in fact
have served to underrnine the position of wornen 111 mdlgenous
society because men are likely to bave received a much greater
share ~f the favours that the Roman state routinely dispensed to
those provincials who collaborated in military al:~ civilian govern-
ment. It is suggestive, perhaps, that the tradIllOnally powerful
position of wornen in Cherokee society:-in tri.bal warfare, govern-
anee and ritual-was seriously eroded In the eJghteenth century by
Eur~pean settlement on the Appalaehian fron tier, . in large part
beeause they were excluded from inter-colomal pohl1cal negol1a-
tions, and therefore denied access to the gifts that routinely accom-
panied them 7 '
7°1. Marion, 'La Population de Volubilis a l'epoque romaine', Bulletin d'archl!o-
logie marocaine, 4 (1960), 175; cf. Brett and Fentress, Berhers, 61; MacMu11en,
Changes, 294-5 n. 3°· .,. D M' 1 11
7 1 T. Hatley, 'Cherokee Wornen Farmers Hold thelf Ground , Ill. R. . l:C le
(ed.), Appalachian Frontiers: Settlement, Society {Im} Dcre!oprnent m the Premdus-
trial Era (Lexington, Mass., T99I), 47·

66 Cf. D. 1. Buck, 'Prontier Processes in Roman Tripolitania', in Buck and Mat-

tingly (eds.), Town and Country, I87; G. Woolf, 'The Formation of Roman Provin-
cial Cultures', in 1. Metzler et al. (eds.), Integration in the Early Roman West: The
Role of Culture and Ideology (Luxembourg. 1995), 9-I8 (on Gaul).
67 1. Slofstra, 'An Anthropological Approach to the Study of Romanization
Processes', in Bralldt and Siofstra (eds.), Roman and Nati)!e, 79, 82.
68 See Whittaker, 'Land and Labour', 333, 343, 350. See also Broughton, Roma-
nization of Africa, 225 (thc 'sodal basis' of north Africa was 'never appreciably
challged').
69 MacMullcn: Changes, 63. See also E. Fantham el al., W0111en in the Classical
World (Oxford and New York, 1994), 374.

6 seems, by Roman practices4
Conclusion: The Limits
0/ Acculturation
In the end, it must be admitted that we do not have, and are
perhaps now never likely to possess, the sort of evidence that
would allow us to define exactly either the extent of Romaniza-
tion in the fron tier-zone Of its effect on the lives of ordinary
north Africans. It has been said recently that the area was 'defi-
nitely Roman' already by the time of the Flavians.' N orth Africa,
we are told, 'imbibed Roman culture', except in Hs mountainous
regions.' It seems to me that the weight of the evidence tells
against the view that the Romans brought about any really
significant change in north African society. Their principal
contribution, it might be said, was to open new markets to north
African products.
It is reasonably clear also, I think, that the social and cultural
consequences of the Roman occupation of north Africa were feIt
mainly in its larger urban centres and among the weaIthy.3 It may
be agreed with H.-G. Pflaum that there were in fact two north
Africas-one that was urbanized (though perhaps not really
urban), relatively and probably increasingly wealthy, and, in
same important aspects of Hs material culture, Romanized;
r P. MacKendrick, The North African Stones Speak (Chapel Hili, Ne, 1980),216.
21. A. Ilevbare, 'Same Aspects of Social Change in North Africa in Punic and
Roman Times', Museum Africum, 2 (r973), 32.
3 Cf. T. R. S. Broughton, The Romanization 0/ Africa Proconsularis (Baitimore,
1929), 140-1; C. R. Whittaker, 'Integration of the Early Roman West: The Example
of Africa', in 1. Metzler el al. (eds.), Integration in the Early Roman West: The Rote
01 Culture and Ideology (Luxembourg, 1995), 19-32. Much the same is said of
Europe in A. T. Fear, Rome and Baetica: Urbanization in Southern Spain c. 50
BC-AlJ 150 (Oxford, 1996), 46, 60, 265; R. MacMullen, Changes in the Roman
Empire: Essays in the Ordinary (Princeton, 1990),42,46; P. 0rsted, Roman Imperial
Economy and Romanization (Copenhagen, 1985), 199.
The Limits 0/ Acculturation 159
another that was rural and mountainous, and mostly untouched, it
There can be little doubt that the local elites were, in some
(indeterminate) measure, Romanized;' they are likely, over time,
to have fashioned a sort of composite culture, in which architec-
tural forms and burial customs, for example, might be imitative of
Roman styles, while the language of everyday discourse was indi-
genous 6 But those who embraced Roman ways were almost
certainly in the minority.7 The poor, after all, really had nothmg
to gain by giving up their customs. lt seems to me that Robert
Broughton probably had it exact1y right when he said that 'the
ways ofthe mass ofthe people never changed,8lt is ironie that the
novelist Louis Bertrand reached much the same conclusion in the
1930s, but on very different grounds; the indigenous population of
Roman-era Algeria, he believed, was morally and intellectually
incapable of assimilating the superior and civilizing tendencies of
its Roman conquerors. 9
Against the many indications of cultural and social continuity in
Roman-era north Africa we are asked to set the example of the so-
called 'Maktar harvester';'" born to an impoverished and unpro-
pertied local family, he worked for years as an itinerant field-hand,
became foreman of a gang of labourers, eventually bought a house
and land, and served on the town-council. His rags-to-riches tale
was undoubtedly replicated, perhaps many times over, in the
course of the Roman occupation. But it must be admitted also
that the vast majority of itinerant field-workers are unlikely to have
finished their careers in municipal office. In so far as the Roman
4 Pflaum: 'La Romanisation de l'Afrique', Vestigia, 17 (1973), 65· Cr. Ilevbare,
'Aspects of Social Change', 39; D. J. Mattingly, 'Libyans and the Limes: Cultu~e and
Society in Roman Tripolitania', Ant. afr. 23 (1987), 84; R. MacMullen, Rural
Romanization', Phoenix, 22 (1968), 337-41.
5 Cf. C. R. Whittaker, 'Land and Labour in North Africa', Klio, 60 (1978), 33 1,
334. (1 MacKendrick, North Alrican Stones, 327·
7 See also MacMullen, Changes, 42.
8 Romanization 01 Africa, 228; cf. 113, 155, 226.
9 Bertrand: Sur les Routes du Sud, 9th edn. (Paris, 1936), 217; see also P. M. E.
Lorcin, Imperial Identities: Stereotyping, Prejudice and Race in Colonial AIgel'ia
(London and New York, 1995), 202.
10 MacKendrick, North Alrican Stones, 76; see also C. M. WeHs, 'The Problems
of Desert Frontiers', in V. A. Maxfield and M. J. Dobson (eds.), Roman Frontier
Studies 1989: Proceedings 01 the XVth International Congress 01 Roman. Frontier
Studies (Exeter, 1991),479. The verse inscription that records hIS career IS CIL 8.
1I284 (~ ILS 7457)·
 160 The Limits 0/ Acculturation
oeeupation may have widened the gap between rieh and poor, it
wIll have served to expand also the range across which upward (or
down ward) mobility eould oeeur. lt is neeessary to insist however
that the mobility evideneed by the 'Maktar harveste/ has ver;
11ttle to do with Romanization: are we to suppose that hard work
normally went unrewarded in indigenous soeiety?
What I have said about Rome's effeet on north Africa is likely
also to have been true of the Algerian fron tier-zone. lt has been
suggested that the western valleys of the Auros mountains were
Romanized." But the only evidence that has ever been adduced to
demonstrate the Romanization of the Aures is an inscription from
Mena'a (eIL 8. 4204) which indicates that there were coloni
('tenant-farmers') there in AD 166-hardly sufficient grounds for
coneluding, with Elizabeth Fentress, that the mountain-valleys
were densely occupied by 'Romanised' settlements. I2 Broughton's
careful reading of the inscriptions led hirn to conelude that the
Auros was a 'strongly indigenous region' throughout the period of
the Roman oeeupation. '3 Even at that, the Aures, and perhaps also
the Nememshas, were probably more thoroughly Romanized than
many of the communities of western Algeria (Mauretania Caesar-
iensis). '4 Paul MacKendrick may be right in thinking that even at
Thamugadi, which, in its architecture and amenities was perhaps
the most eonspicuously Roman of all the settlements of the fron-
tier-zone, Romanization was no more than a 'thin veneer' . I 5
In the almost complete absence of evidence for any significant
measure of cultural change in the region, it is difficult now to
understand why so much modern writing about north Africa has
J I ,E. W. B. Fe~tress, Numidia and the Roman Army: Social, Military and Eco-
nomlC Aspects o} (he Frontier Zone (Oxford, I979), 143' see also Y. Le Bohec La
Troisieme Legion Auguste (Paris, 1989),578.' , counted on the Romans to defend their interests, the indigenous
• 12 F~ntress: 'Forever Berber?', Opus, 21I (1983), 167. There is no basis for believ-
lOg, ,,:,lth Fentress, ibid. 166, P. A. Fevrier, 'Il1scriptions inedites relatives aux
d?~al~eS de la ~egion.de ~etif', i.n R. Chevallier (ed.), Melanges d'archeologie et
d 11lstOire o//erts a A. PlganlOl (Pans, 1966), 215, that all the coloni attested on north
Afr~can inscriptions were tenants of imperial estates; see B. D. Shaw, 'Soldiers and
SO~lety: Th~ A~my in. Nu~idia', Opus, 2/~ .(1983), I4I.
. RomamzatlOn 0/ .Af~/ca, IJ9 n. 4 (cItmg also Procopius, Vand. 4. 13. 23-5,
WhlCh seems to me to mdlCate only that the mountains were rugged and difficult of
passage); cf: 139 ('native life continued largely untouched in the mountains'). See
als1~ P. Monzot, 'Vues nouyelles ~ur l'Aures antique', CRAl (1979), 309-10.
.Cf. R. I. Lawless, 'L'EvolutlOll du peuplement, de l'habitat et des paysages
ag~aJres du Maghreb', Annales de geographie, 8r (1972),460.
5 North African Stones, 240.
The Limits 0/ Acculturation 161
maintained that it was extensively Romanized. I suppose that some
part of the explanation is to be found in once widely shared,
mostly implicit assumptions about the superiority of European
civilization. lt is perhaps also a product of the seemingly widely
held belief that the army functioned both as a powerful instrument
of Romanization in the provinees (an idea which, it seems to me,
may be in need of some revision), and as an intermediary betw~en
the intrusive and indigenous cultures. There is no reason to thmk
that the army performed either role in the Algerian fron tier-zone. ,6
The epitaphs examined earlier (Chapter 4) would seem to indicate
that very few soldiers married un-Romanized women in any part of
the frontier-zone. lt was remarked earlier also that there IS 11ttle
evidence of any other kind of cross-cultural interrelationship.
Recent study that characterizes the Roman frontier-regions as
zones of cultural interpenetration cannot readily accommodate
. '7
the experience of Roman-era A 1germ.
We might therefore want to imagine that there was in fact a third
north Afriea, one of forts and fortifications, administrators and
soldiers who were increasingly of African origin, but segregated,
by eirc~mstance or by choiee, from the people they ruled; their
own society (and solitude), as it were, with little discernible effect
on the structures and rhythms of north African life. Brent Shaw
has suggested that the soldiers stationed in the region (whose
deployment seems to have been intended at least as much to
tarne the local population as to protect it) were removed also
from the concerns of its inhabitants, identifying themselves instead
'wholly with the interests of their commanding officers and of a
militaristic central government'. r8 I wonder whether we mlght not
also suppose that, apart from the propertied elites, who will have
peoples of the frontier-zone were no less estranged from the army
that occupied their land.
16 Cf. Fentress, Numidia, 156; Shaw, 'Soldiers and Society', I51: the. army 'was
not halfway between the ruler and the ruled; it was the instrument of viOlent force
wielded by the central power structure of the empire' (his italics).
17 'Recent study'; e.g. S. L. Dyson, The Creation of the Roman Frontfer (Prince-
ton, 1985),4; W. S. Hanson, 'The Nature and Function ofRoman Frontiers:' in 1. C.
Barrett, A. P. Fitzpatrick, and L. Macinnes (eds.), Barbarians and Romans 111 North-
West Europe: From the Later Republic to Late Antiquity (Oxford, 19 89), 55·
18 'Soldiers and Society', 148.

Bibliography of Modern Works Cited
in the Notes
Abdelalirn, M. K., 'Libyan Nationalism and Foreign Rule in Graeco-
Roman Times' , in Libya Antiqua: Report and Papers of the Sy'mposium
Organized by UNESCO in Paris, /8 to /9 January, /984 (Pans, I986),
I53--63. . 'Acculturation', American Anthropologist, 56 (1954), 973-1000.
Adams, J. N., 'Latin and Punic in Contact? The Case of the Bu NJem
Ostraca', JRS 84 (I994), 87-II2.
Alexander, J. A., 'Prontier Studies and the Earliest Farmers in Europe' , in
D. Green, C. Haselgrove, and M. Spriggs (eds.), Social Organisation and
Settlement: Contributions from Anthropology, Archaeology, and Geogra-
phy (Oxford, I978), I3-29.
- - 'Early Frontiers in Southern Africa', in M. Hall et al. (eds.), Fron-
tiers: Southern African Archaeology Today (Oxford, 1984), 12-2 3.
- 'The End of the Moving Frontier in the Neolithic of North-Eastern
Africa', in L. Krzyzaniak and M. Kubosiewicz (eds.), Origin and Ea~ly
Development 0/ Food~Producing Cultures in North-Eastern A[nca
(Poznan, I984), 57-63. . . ,
Alföldy, G., 'Notes sur la relation entre le drOlt de cIte et la nomenc1ature
dans l'empire romain', Latomus, 25 (1966), 37-57.
Allanson-Jones, L., 'Roman and Native Interaction in Northumberland',
in V. A. Maxfield and M. 1. Dobson (eds.), Roman Frontier Studies 19 89:
Proceedings o[ the XVth International Congress o[ Roman Frontier
Studies (Exeter, I99I), I-5. . " . historische Quelle (Cologne and Vienna, I986), 1 I6-66.
Aquilella Almer, 1., and Lopez Cerda, ~. A." 'Deter~ll~atlO~ d~ la r.epr~~
sentativite des inscriptions latines grace a Ja statIstlque mferentIelle,
Ant. afr. 9 (I975), II5-26. . .
Arangio~Ruiz, v., 'L'origine deI testamenturn militis e la sua poslzlOne nel
diritto romano classico', Bullettino dei Istituto di Diritto Romano, 18
(I906), I57-9 6 . . .
Ballhatchet, K., Race, Sex and Class under the Raj: Impenal Attztudes and
Policies and their Critics, 1793-1905 (New York, 1980).
Balsdon, J. P. V. D., Romans and AUens (London, I979).
Baradez, 1.., Fossatum Africae: Recherehes aeriennes sur l'organisation des
a
conjins sahariens l'epoque romaine (Paris, 1949).
--- 'Complements inedits au Fossaturn Africae', in Studien zu den
Bibliography 255
Militärgrenzen Roms: Vorträge des 6 internationalen Limeskongresses in
Süd-Deutschland (Cologne, I966), 200-IO.
Barker, G. W. w., 'The UNESCO Libyan Valleys Survey: Developing
MethodoJogies for Investigating Ancient Floodwater Farming', in D.
1. Buck and D. 1. Mattingly (eds.), Town and Country in Roman Tripo~
litania: Papers in Honour of Olwen Hacketl (Oxford, I985), 29 I -3 0 6.
- - and Jones, G. D. B., 'The UNESCO Libyan Valleys Survey III,
Palaeo~Economy and Environmental Archaeology in the Pre~Desert',
Lib. Stud. I3 (I982), I-34.
- - 'The UNESCO Libyan Valleys Survey VI: Investigations of a
Romano-Libyan Farm', üb. Stud. 15 (I984), I-44.
Barnett, H. G., Broom, L., Siegel, B. 1., Vogt, E. Z., and Watson, 1. B.,
Barrett, 1. c., 'Afterword: Render Vnto Caesar', in 1. C. Barrett, A. P.
Fitzpatrick, and L. Macinnes (eds.), Barbarians and Romans in North-
West Europe: From the Later Repuhlic to Late Antiquity (Oxford, 19 89),
235-4I.
- - Fitzpatrick, A. P, and Macinnes, L. (eds.), Barbarians and Romans in
North~ West Europe: From the Later Republic to Late Antiquity (Oxford,
I9 89).
BarteI, B., 'Colonialism and Cultural Responses: Problems Related to
Roman Provincial Analysis', W. Arch. I2/I (1980), II-26.
- - 'Comparative Historical Archaeology and Archaeological Theory',
in S. L. Dyson (ed.), Comparative Studies in the Archaeology o[ Coloni-
aUsm (Oxford, I985), 8-37.
Bee, R. L., Patterns and Processes: An Introduction to Anthropological
Strategies for the Study of Sociocultural Change (New York, I974).
Behrends, 0., 'Die Rechtsregelungen der Militärdiplome und das die
Soldaten des Prinzipats treffende Eheverbot' , in WEck and H. Wolff
(eds.), Heer und Integrationspolitik: Die römischen Militärdiplome als
Benabou, M., La Resistance africaine cl la romanisation (Paris, I97 6).
- - 'Les Romains ont-ils conquis I'Afrique?', Ann. ESC 33 (I978), 83-8.
Benseddik, N., Les Troupes auxiliaires de I' armee romaine en Mauretanie
Cesarienne sous le Haut~Empire (Algiers, 1982).
Berkhofer, Jr., R. F., 'The North American Frontier as Process and
Context', in H. Lamar and L. Thompson (eds.), The Frontier in History:
North America and Southern Africa Compared (New Haven and
London, I98I), 43-75.
Berthier, A., 'Nicibes et Suburbures: Nomades ou sedentaires?', Bulletin
d'archeologie algerienne, 3 (I968), 293-3 00 .
- - La Numidie: Rome et la Maghreb (Paris, 1981 ).
Bertrand, L., Sur les Routes du Sud, 9th edn. (Paris, I936).
 Bibliography
Birebent, 1., Aquae Romanae: Recherehes d'hydraulique romaine dans fest
algerien (Algiers, 1964).
Birley, A. R., 'Roman Frontiers and Roman Frontier Policy: Some Reftec-
t1003 on Roman Imperialism', Transactions 0/ the Architectural and
Archaeological Association 0/ Durham and Northumberland, 3 (I974),
13-25.
- - 'The Economic Effects of Roman Frontier Policy', in A. C. King and
M. Henig (eds.), Military and Civilian in Roman Britain (Oxford, I9 8r ),
39-54·
Birley, E., 'Hadrianic Frontier Policy', in E. Swoboda el al. (eds.),
Carnuntina: Ergebnisse der Forschung über die Grenzprovinzen des
römischen Reiches: Vorträge beim internationalen Kongress der Alter-
tumsforscher Carnuntum 1955 (Graz and Cologne, 1956), 25-33.
- - 'Before Diplomas, and the Claudian Reform', in WEck and H.
Wolff (eds.), Heer und Integrationspolitik: Die römischen Militärdiplome
als historische Quelle (Cologne and Vienna, 1986), 249-57.
Blagg, T. F. C., 'Art and Architecture', in M. Todd (ed.), Research on
Roman Britain [960-89 (London, 1989), 203-17.
- - 'First-Century Roman Houses in Gaul and Britain', in T. F. C. Blagg
and M. Millett (eds.), The Early Roman Empire in the West (Oxford,
1990), 194-209.
- - and Millett, M., 'Introduction', in T. F. C. Blagg and M. Millett
(eds.), The Early Roman Empire in the West (Oxford, 1990), 1-4.
- - (eds.), The Early Roman Empire in the West (Oxford, 1990).
Bloemers, J. H. F., 'Acculturation in the Rhine/Meuse Basin in the Roman
Period: A Preliminary Survey', in R. Brandt and J. Siofstra (eds.),
Roman and Native in the Low Countries: Spheres of Interaction (Oxford,
1983), 159-20 9.
- - 'Aeeulturation in the Rhine/Meuse Basin in the Roman Period:
Demographie Considerations', in J. C. Barrett, A. P. Fitzpatrick, and
L. Macinnes (eds.), Barbarians and Romans in North- West Europe: From
the Later Republic to Lote Antiquity (Oxford, 1989), 175-97.
- - 'Lower Germany: Plura consilio quam vi: Proto-Urban Settlement
Developments and the Integration of Native Cornmunities', in T. F. C.
Blagg and M. Millett (eds.), The Early Roman Empire in the West
(Oxford, 1990), 72-86.
- - 'Relations between Romans and Natives: Concepts of Comparative
Studies', in V A. Maxfield and M. J. Dobson (eds.), Roman Frontier
Studies 1989: Proceedings of the XVth International Congress of Roman
Frontier Studies (Exeter, 1991),451-4.
Böhm, R., 'Zur lex Minicia', Zeitschrift der Savigny-Stiftung für
Rechtsgeschichte (Romanistische Abteilung), 84 (1967),363-71.
Bibliography 257
Bouchenaki, M., 'La Recherche areheologique en Algerie', in H. G. Horn
and C. B. Rüger (eds.), Die Numida' Reiter und Könige nördlich der
Sahara (Cologne, 1979), 1-3.
--, 'Recentes recherehes et etude de l'antiquite en Algerie', Ant. afr. 15
(1980),9-28.
Bradley, R., 'Foreword: Trial and Error in the Study of Mortuary
Practices-Exploring the Regional Dimension', in L. A. Beck (ed.),
Regional Approaches to Mortuary Analysis (New York and London,
1995), pp. v-ix.
Brandt, R., 'A Brief Encounter along the Northern Frontier', in R. Brandt
and J. Slofstra (eds.), Roman and Native in the Low Countries: Spheres 0/
Interaction (Oxford, 1983), 12 9-45.
--- and Slofstra, J. (eds.), Roman and Native in the Low Countries:
Spheres of Interaction (Oxford, 19 83).
Branigan, K., 'Celtic Farm to Roman Villa', in D. Miles (ed.), The
Romano-British Countryside: Studies in Rural Settlement and Economy
(Oxford, 1982), i. 81-96.
Braund, D., 'Introduction: The Growth of the Roman Empire (241 BC-AD
193)', in D. Braund (ed.), The Administration ofthe Roman Empire (24 1
BC-AD [93) (Exeter, 1988), 1-13.
- - 'Ideology, Subsidies and Trade: The King on the Northern Frontier
Revisited', in J. C. Barrett, A. P. Fitzpatriek, and L. Macinnes (eds.),
Barbarians and Romans in North- West Europe: From the Later Republic
to Late Antiquity (Oxford, 1989), 14-26.
Breeze, D. J., 'Demand and Supply on the Northern Frontier', in R. Miket
and C. Burgess (eds.), Between and beyond the Walls: Essays on the
Prehistory and History 0/ North Britain in Honour of George Jobey
(Edinburgh, 1984), 148-65.
- - 'The Impact ofthe Roman Army on North Britain', in J. C. Barrett,
A. P Fitzpatrick, and L. Macinnes (eds.), Barbarians and Romans in
North- West Europe: From the Later Republic to Late Antiquity (Oxford,
19 89), 227-34·
- - 'The Northern Frontiers', in M. Todd (ed.), Research on Roman
Britain [960-89 (London, 1989), 37-60.
- - and Dobson, B., Hadrian's Wall (London, 197 6).
Brett, M., and Fentress, E. W B., The Berbers (Oxford, 1996).
Brogan, 0., 'Hadd Hajar, a "Clausura" in the Tripolitanian Gebel Garian
South of Asabaa', Lib. Stud. 11 (19 80),45-5 2.
Broughton, T. R. S., The Romanization ofAfrica Proconsularis (Baltimore,
19 29).
Brown, J., 'On Mortuary Analysis-with Special Reference to the Saxe-
Binford Research Program', in L. A. Beck (ed.), Regional Approaches to
Mortuary Analysis (New York and Lendon, 1995), 3-2 6.
 Bibliography
Brunt, P. A., Italian Manpower 225 B.C-A.D. /4 (Oxford, 1971).
- - 'The Romanization of the Local Ruling Classes in the Roman
Empire', in D. M. Pippidi (ed.), Assimilation el resistance ci la culture
greco-romaine dans le monde ancien: Travaux de Vle Congres Interna-
tional d'Etudes Classiques (Madrid, Sept. 1974) (Bucharest and Paris,
1976), 161-73·
Buck, D. J., 'Prontier Processes in Roman Tripolitania', in D. J. Buck and
D. J. Mattingly (eds.), Town and Country in Roman Tripolitania: Papers
in Honour oJ Olwen Hackett (Oxford, 1985), 179-90.
Buckland, W W" A Textbook 0/ Roman Law from Augustus to Justinian,
3rd edn., rev. by P. Stein (Cambridge, 1966).
Buren, A. W van, 'Some Families Formed by Roman Soldiers and Veter-
ans', in M. Renard (ed.), Hommages a Albert Grenier (Brussels, 1962),
15 64-7°.
Burnham, R c., 'Pre-Roman and Romano-British Urbanism? Problems
and Possibilities', in B. C. Burnham and H. R Johnson (eds.), Invasion
and Response: The Case of Roman Britain (Oxford, 1979), 255-72.
- - 'The Origins of Romano-British Small Towns', Oxlord Journal of
Archaeology, 5 (1986), 185-2°5.
Bums, 1. R., and Denness, B., 'Climate and Social Dynamics: The Tripo-
litanian Example, 300 BC-AD 300', in D. 1. Buck and D. 1. Mattingly
(eds.), Town and Country in Roman Tripolitania: Papers in Honour of
Olwen Hackett (Oxford, 1985), 201-25.
Cadenat, P., 'Un etablissement pre-romain dans la region de Tiaret (Ora-
nie)', An!. afr. 6 (1972), 29-58.
Cagnat, R., L'Armee romaine d'Afrique et l'oeeupation militaire de I'Afri-
que sous les emperfurs, 2nd edn. (Paris, 1913).
- - Merlin, A., and Chatelain, L. (eds.), Inscriptions latines d'Alrique
(Tripolilaine, Tunisie, Maroe) (Paris, 1923).
Campbell, B., 'The Marriage of Soldiers under the Empire', JRS 68
(1978), 153-66.
- - The Emperor and the Roman Army, 31 Be-AD 235 (Oxford, 1984).
Camps, G., 'L'Inscription de Beja et le probleme des Dii Mauri', Revue
africaine, 97 (1954), 233-60.
- - 'Aux origines de la Berberie: Masinissa ou les d6buts de l'histoire',
Ubyea, 8 (1960), 1-320.
--·Les Civilisations prehistoriques de I'Afrique du Nord et du Sahara
(Paris, 1974).
- - 'Qui sont les Dii MauriT, Ant. afr. 26 (1990), 131-53.
Camps-Farber, H., L'Olivier el l'huile dans I'Afrique romaine (Algiers, 1953).
Carandini, A., 'Pottery and the African Economy', in P. Garnsey, K.
Hopkins, and C. R. Whittaker (eds.), Trade in the Ancient Economy
(Berkeley and Los Angeles, 1983), 145-62.
Bibliography 259
CasteIlo, c., 'La data della legge Minicia', in Studi in onore di V Arangio-
Ruiz (NapIes, 1953), iii. 301-I7.
Chabot,1. R, Recueil des inscriptions libyques (Paris, 1940).
Cherry, D., 'The Minician Law: Marriage and the Roman Citizenship'
Phoenix, 44 (1990), 244-66. '
- - 'Re-figuring the Roman Epigraphic Habit', The Ancient History
Bulletin, 9 (1995), 143-56.
- - 'Marriage and Acculturation in Roman Algeria', Classieal Philology,
9 2 (r997), 71-83.
Chevallier, ~., 'La Centuriation et les problemes de la colonisation
romaine', Etudes rurales, 3 (1961), 54-80.
Christol, M., 'Rome et les tribus indigenes en MaurManie Tingitane'
L'Ajrica romana, 5 (1988), 305-37. '
Claessen, H. 1. M., 'Kinship, Chiefdom and Reciprocity: On the Use of
Anthropological Concepts in Archaeology', in R. Brandt and 1. Slofstra
(eds.), Roman and Native in the Low Countries: Spheres of Interaction
(Oxford, 1983),211-22.
Clarke, G., 'ULVS XIV: Archaeozoological Evidence for Stock-Raising
and Stock-Management in the Pre-Desert', Lib. Stud. 17 (1986), 49-64.
Cooter, WS., 'Preindustrial Frontiers and Interaction Spheres: Prolego-
menon to a Study of Roman Frontier Regions', in D. H. Miller and 1. 0.
Steffen (eds.), The Frontier: Comparative Studies (Norman, Okla.,
1977), 81-107.
Crawford, D. 1., 'Imperial Estates', in M. 1. Finley (ed.), Studies in Roman
Property (Cambridge, 1976), 35-7 0 .
Craok, 1., Law and Life oJ Rome (London, 19 67).
Daniels, c., The Garamantes 01 Southern Libya (Stoughton, Wis., 1970).
- - 'The Garamantes ofFezzan', in F. F. Gadallah (ed.), Libya in History
(Benghazi, 1971), 261-85.
- - 'The Frontiers: Africa', in 1. S. Wacher (ed.), The Roman World
(London and New York, 1987), i. 223-65.
Darmon, 1.-P., 'Note sur le tarif de Zarai', Cahiers de Tunisie, 12 (19 64),
7-2 3.
Daube, D., Roman Law: Linguistic, Social and Philosophical Aspects
(Edinburgh, 1969).
Davies,1. L., 'Soldiers, Peasants and Markets in Wales and the Marches'
in T. F. C. Blagg and A. C. King (eds.), Military and Civilian in Roma~
Britain: Cultural Relationships in a Frontier Province (Oxford, 19 84),
93- 12 7.
Davies, R. W., Service in the Roman Army, ed. D. Breeze and V. A.
Maxfield (New York, 1989).
DeneIl, R., 'The Hunter-Gatherer/Agricultural Frontier in Prehistoric
Temperate Europe', in S. W Green and S. M. Perlman (eds.), The
Archaeology of Frontiers and Boundaries (Orlando, Fla., 1985), 113-4 0 .
 260 Bibliography
Desanges, J., Catalogue des tribus africaines de l' antiquite classique ci
l'Ouest du Ni! (Dakar, 1962),
D'Escurac Doisy, H., 'Lambese et les veterans de la legio tertia Augusta',
in M. Renard (ed.), Hommages ci Albert Grenier (Brussels, 1962), 571-83.
- - 'Le Phenomene associatif dans le monde paysan al'epoque du Haut~
Empire', An!, afr. 1 (1967), 59-71.
Despois, J., Le Hodna (Algerie) (Paris, 1953). . .
- - La Tunisie orientale, Sahel ef basse steppe, 2nd edn. (Pans, 1955).
- - 'La Culture en terrasses en Afrique du Nord', Ann. ESC 2 (1956),
42-5 0 . .
--- Histoü'e de l'utilisation des terres des regions arMes (Pans, 1961).
- - and Raynal, R., Geographie de l'AJrique du Nord-Ouest (Paris, 1967).
Dobson, B., 'The Roman Army: Wartime Of Peacetime Army', in WEck
and H. Wolff (eds.), Heer und Integrationspolitik: Die römischen Mili-
tärdiplome als histor;sche Quelle (Cologne and Vienna, 1986), 10-25.
Dohrenwend, B. P., and Smith, R. 1., 'Toward a Theory of Acculturation',
Southwestern Journal or Anthropology, 19 (1962), 30-9.
Dommelen, P van, 'Colo~ial Constructs: Colonialism and Archaeology in
the Mediterranean', WAreh. 28/3 (1997), 305-23.
Dondin-Payre, M., 'Recherehes sm un aspect de la romanisation de
l'Afrique du Nord: L'Expansion de la citoyennete romaine jusqu'a
Hadrien', Ant. aJr. 17 (1981), 93-132.
Driel-Murray, C. van, 'Gender in Question', in P. Rush (ed.), Theoretical
Roman Archaeology: Second Conjerence Proceedings (Aldershot and
Brookfield, Vt., 1995), 3-21.
Drinkwater, 1. F., 'For Better or Worse? Towards an Assessment of the
Economic and Sodal Consequences of the Roman Conquest of Gaul',
in T. F. C. Blagg and M. Millett (eds.), The Eorly Roman Empire in the
West (Oxford, 199°),210-19.
Drummond, S. K., and Nelson, L. H., The Western Frontiers 0/ Imperial
Rome (Armonk and London, 1994).
Duncan-Jones, R. P, 'Wealth and Munificence in Roman Africa', Papers
oJ the British School at Rome, 31 (1963), 159-77.
- 'Some Configurations of Landholding in the Roman Empire', in
M. I. Finley (ed.), Studies in Roman Property (Cambridge, 1976), 7-34.
- The Economy 0/ the Roman Empire: Quantitative Studies, 2nd edn.
(Cambridge, 1982). . Fevrier, P. A., 'Inscriptions inedites relatives aux domaines de la region de
Dupuis, X., 'La Participation des veterans ala vie municipale en Numidle
meridionale aux He et IHe siecles', in C. Lepelley (ed.), Actes du IVe
colloque international sur l'histoire el l'archeologie de I'Afrique du Nord
(Strasbourg, 5-9 avri!, I988) (Paris, 1991), 343-54.
Dyson, S. L., The Creation 0/ the Roman Frontier (Princeton, 1985). .
Eadie, J., 'Peripheral Vision in Roman History' , in 1. D' Arms and J. Eadle
Bibliography 261
(eds.), Ancient and Modern Essays in Honor 0/ Gerald F Else (Ann
Arbor, 1977),215-34.
- - 'Civitates and Clients: Roman Frontier Policies in Pannonia and
Mauretania Tingitana', in D. H. Miller and 1. 0. Steffen (eds.), The
Frontier: Comparative Studies (Norman, Okla., 1977), 57-80.
Eck, W, and Wolff, H., 'Ein Auxiliardiplom ans dem Jahre 203 n. Chr.',
in WEck and H. Wolff (eds.), Heer und Integrationspolitik: Die
römischen Militärdiplome als historische Quelle (Cologne and Vienna,
19 86), 556-75.
Edmondson, J. c., 'Romanization and Urban Development in Lusitania',
in T. F. C. Blagg and M. Millett (eds.), The Early Roman Empire in the
West (Oxford, 1990), 151-78.
Elmayer, A. F., 'The Reinterpretation of Latino-Punic Inscriptions from
Roman Tripolitania', Lib. Stud. 14 (1983),86-95, and 15 (1984), 93- 10 5.
Elton, H., Frontiers 0/ the Roman Empire (Bloomington, Ind., 1996).
Euzennat, M., 'Quatre annees de recherches sur la frontiere romaine en
Tunisie meridionale', CRAI (1972), 7-27.
- - 'Les Recherehes sur la frontiere romaine d'Afrique', in 1. Fitz (ed.),
Akten des XI internationalen Limeskongresses, Szekes/ehervar, 1976
(Budapest, 1977),533-43.
- - 'L'Olivier et le "Limes", considerations sur Ia frontil~re romaine de
Tripolitanie', BGTH (1985), 161-71.
- - Le Limes de Tingitane: La Frontiere meridionale (Paris, 19 89).
- - 'La Frontiere romaine d' Afrique', GRAl (1990), 565-80.
Fantharn, E., Foley, H. P., Kampen, N. B., Pomeroy, S. B., and Shapiro,
H. A., Women in the Glassical World (Oxford and New York, 1994).
Fear, A. T., Rome and Baetica: Urbanization in Southern Spain c. 50 BC-AD
I50 (Oxford, 1996).
Fentress, E. W B., 'Dii Mauri and Dii Patrii', Latomus, 27 (197 8),5 0 7- 16.
- - Numidia and the Roman Army: Social, Military and Economic
Aspects 0/ the Frontier Zone (Oxford, 1979).
- - 'Forever Berber?', Opus, 2/r (1983), 161-75.
Ferris, I., 'Shopper's Paradise: Consumers in Roman Britain', in P Rush
(ed.), Theoretical Roman Archaeology: Second Con/erence Proceedings
(Aldershot and Brookfield, Vt., 1995), 13 2-4 0.
Setif', in R. Chevallier (ed.), Midanges d'archeologie et d'histoire offerts
ci A. Piganiol (Paris, 1966), 207-28.
- - 'Religion et domination dans l' Afrique romaine', Dialogues d'histoire
anäenne, 2 (1976), 305-36.
- - Approches du Maghreb romain, 2 vols. (Aix-en-Provence, 19 8 9-9 0 ).
- - and Guery, R., 'Les Rites funeraires de Ia necropole orientale de
Setif', Ant. aJr. 15 (1980), 9 1- 12 4.
262 Bibliography
Fink, R. 0., 'Hunt's Pridianum: British Museum Papyrus 2851', JRS 48
(1958), 102-16.
-- Roman Military Records on Papyrus (Cleveland, 1971).
Fishwick, D., 'TheAnnexation ofMauretania', Historia, 20 (1971), 467-87.
Fitzpatrick, A. P., 'The Uses of Roman Imperialism by the Celtic Bar-
barians in the Later Republic', in J. C. Barrett, A. P. Fitzpatrick, and
L. Macinnes (eds.), Barbarians and Romans in North- West Europe:
From the Later Republic to Late Antiquity (Oxford, 1989), 27-54-
Forni, G., Il reclutamento delle legion i da Augusto a Diocleziano (Milan,
1953)·
- - 'I diplorni militari dei c1assiari delle flotte pretorie (inclusi quelli dei
c1assiari-Iegionari)', in WEck and H. Wolff (eds.), Heer und Integra-
tionspolitik: Die römischen Militärdiplome als historische Quelle
(Cologne and Vienna, 1986), 293-321.
Freeman, P. W M., '''Romanisation'' and Roman Material Culture', JRA
6 (1993), 438-45.
Frezouls, E., 'Rome et la Mauretanie Tingitane: Un constat d'6chec?', Ant.
afr. 16 (1980), 65-93.
- - 'Les Survivances indigenes dans l'onomastique africaine', in A.
Mastino (ed.), L'Africa Romana: Atti dei VII convegno di studio, Sassari
(Sassari, 1990), 161-6.
Fulford, M., 'The Economy ofRoman Britain', in M. Todd (ed.), Research
on Roman Britain /960-89 (London, 1989), 175-201.
-- 'Roman and Barbarian: The Economy of Roman Frontier Systems',
in 1. C. Barrett, A. P. Fitzpatrick, and L. Macinnes (eds.), Barbarians
and Romans in North~ West Europe: From the Later Republic to Late
Antiquity (Oxford, 1989), 81-95.
Galand, L., 'Les Alphabets libyques', Ant. ajY. 25 (1989), 69-81.
Gardner,1. F., Women in Roman Law and Society (London, 1986).
Garnsey, P., Social Status and Legal Privilege in the Roman Empire
(Oxford, 1970).
-~ 'Septimius Severus and the Marriage of Soldiers', California Studies
in Classical Antiquity, 3 (1970), 45-53.
-- 'Rome's African Empire under the Principate', in P. Garnsey and C.
R. Whittaker (eds.), Imperialism in the Ancient World (Cambridge,
197 8), 223-54, 343-54·
- - Famine and Food Supply in the Graeco~Roman World (Cambridge,
1988).
Gilliam, 1. F., 'Romanization of the Greek East: The Role of the Army',
Bulletin ofthe American Society of Papyrologists, 2 (1965), 65-73; also in
1. F. Gilliam, Roman Army Papers (Amsterdam, 1986), 281-7.
---- 'Dura Rosters and the Constitutio Antoniniana', Historia, 14 (1965),
74-9 2.
Bibliography
Gonzalez, 1., 'The Lex Irnitana: A New Copy of the Flavian Municipal
Law', JRS 76 (1986), 147-243.
Goodchild, R. G., 'Where Archaeology and Military Training Go Hand in
Hand: Roman "Horne Guard" Outposts in Tripolitania', Reports and
Monographs of the Department of Antiquities in Tripolitania, 2 (1949),
29-35·
- - 'The Latino~ Libyan Inscriptions of Roman Tripolitania', Antiquaries
Journal, 30 (1950), 135-44.
- - 'The Limes Tripolitonus', JRS 40 (1950), 30-8; also in R. G.
Goodchild, Libyan Studies: Select Papers of the Lote R. G. G.oodchild,
ed. 1. M. Reynolds (London, 1976), 35-45.
- - 'Oasis Forts of Legio III Augusta on the Routes to the Fezzan',
Papers of the British School at Rome, 22 (1954), 56-65.
- - 'The Roman Roads of Libya and their Milestones', in F. F. Gadallah
(ed.), Libya in History (Benghazi, 1971), 155-71.
Graf, D. F., Review of A. Gutsfeld, Römische Herrschaft und einheimischer
Widerstand in Nordafrika: Militärische Auseinandersetzungen Roms mit
den Nomaden (Stuttgart, 1989), JRS 82 (r992), 277-9.
Green, S. W, 'The Agricultural Colonization of Temperate Forest Habi~
tats: An Ecological Model', in W W Savage and S. l. Thompson
(eds.), The Frontier: Comparative Studies, ii (Norman, Okla., 1979),
69-10 3.
Greene, K., 'Invasion and Response: Pottery and the Roman Army', in
B. C. Burnham and H. B. Johnson (eds.), Invasion and Response: The
Case of Roman Britain (Oxford, 1979), 99-106.
Groenman~van Waateringe, W., 'Urbanization and the North~West Fron-
tier of the Roman Empire', in W S. Hanson and L. 1. F. Keppie (eds.),
Roman Frontier Studies /979 (Oxford, 1980), 1037-44.
- - 'The Disastrous Effect of the Roman üccupation', in R. Brandt and
1. Slofstra (eds.), Roman and Native in the Low Countries: Spheres of
Interaetion (Oxford, 1983), 147-57.
- - 'Food for Soldiers, Food for Thought', in 1. C. Barrett, A. P.
Fitzpatrick, and L. Macinnes (eds.), Barbarians and Romans in
North~ West Europe: From the Later Republic to Late Antiquity
(Oxford, 1989), 96-107.
Gsell, S., Atlas archeologique de tAlgerie (Paris and Algiers, I9 02-II).
- - Khamissa (Paris and Algiers, 1912).
- - Histoire ancienne de tAfrique du Nord, 8 vols. (Paris, 19 20- 8).
Guey, 1., 'Fouilles sur Ie limes romain de Numidie (dans la region de Bordj
Saäda, en fevrier-mars, 1938)" CRAI (1938), 357-9.
Gutsfeld, A., Römische Herrschaft und einheimischer Widerstand in Nord~
afrika: Militärische Auseinandersetzungen Roms mit den Nomaden
(Stuttgart, 1989).
Hauson, WS., 'Administration, Urbanisation and Acculturation iu the
 Bibliography
Roman West', in D. Braund (ed.), The Administration 0/ the Roman
Empire (241 BC-AD 193) (Exeter, 1988), 53-68.
Hanson, WS., 'The Nature and Function of Roman Frontiers', in 1. C.
Barrett, A. P. Fitzpatrick, and L. Macinnes (eds.), Barbarians and
Romans in North- West Europe: From the Later Republic to Late Anti-
quity (Oxford, 1989), 55-63.
Haselgrove, c., 'Culture Process on the Periphery: Belgic Gaul and Rome
during the Late Republic and Early Empire', in M. Rowlands, M.
Larsen, and K. Kristiansen (eds.), Centre and Periphery in the Ancient
World (Cambridge, 1987), 104-24.
- 'The Later Iron Age in Southern Britain and Beyond', in M. Todd
(ed.), Research on Roman Britain 1960-89 (London 1989), 1-18. .
- 'The Romanization ofBelgic Gaul', in T. F. C. Blagg and M. MIlIett
(eds.), The Early Roman Empire in the West (Oxford, 1990),45-71.
Hatley, T., 'Cherokee Women Farmers Hold their Ground', in R. D.
Mitchell (ed.), Appalachian Frontiers: Settlement, Society and Develop-
ment in the Preindustrial Era (Lexington, Mass., 1991), 37-51.
Hedeager, L., 'A Quantitative Analysis ofRoman Imports in Europe North
of the Limes (0-400 A.D.), and the Question of Roman-Germanic
Exchange', in K. Kristiansen and C. Paluden-Müller (eds.), New Direc-
tions in Scandinavian Archaeology (Studies in Scandinavian Prehistory
and Early History, I; Copenhagen, 1978), 191-216.
-- 'Empire, Frontier and the Barbarian Hinterland: Rome and Northern
Europe from AD 1-400', in M. Rowlands, M. Larsen, and K. Kristiansen
(eds.), Centre and Periphery in the Ancient World (Cambridge, 19 87),
12 5-40.
Henig, M., 'Religion in Roman Britain', in M. Todd (ed.), Research on
Raman Britain /960-89 (London, 1989),219-34.
Higham, N. J., 'Roman and Native in England North of the Tees: Accul-
turation and its Limitations' , in J. C. Barrett, A. P. Fitzpatrick, and L.
Macinnes (eds.), Barbarians and Romans in North- West Europe: From
Ihe LaterRepublic 10 LaIe Anliquity (Oxford, 1989), 153-74.
Hingley, R., 'Roman Britain: The Structure of Roman Imperialism and
the Consequences of Imperialism on the Development of a Peripheral
Province', in D. Miles (ed.), The Romano-British Countryside: Studies in
Rural Settlement and Economy (Oxford, 1982), i. 17-52.
-- 'Attitudes to Roman Imperialism', in E. Seott (ed.), Proceedings 01 the
First Theoretical Roman Archaeology Conlerence (Oxford, 1993), 23-7.
- - 'The "Legaey" of Rome: The Rise, Decline and Fall of Romaniza-
tion', in J. Webster and N. Cooper (eds.), Roman Imperialism: Post-
Colonial Perspectives (Leicester, 1996}, 35-48.
Hitchner, R. B., 'The Kasserine Archaeological Survey, 1982-1986', Ani.
afr. 24 (1988), 7-41.
-- 'The Kasserine Archaeological Survey, 1987', Anl. afr. 26 (1990),23 I-59.
Bibliography 26 5
Hodder, I., 'Pre-Roman and Romano-British Tribal Economies', in B. C.
Burnham and H. B. Johnson (eds.), Invasion and Response: The Case 01
Roman Britain (Oxford, 1979), 18 9-9 6.
- - and Millett, M., 'Romano-British Villas and Towns: A Systematic
Analysis', WAreh. 12 (1980), 69-76.
Hogg, A. H. A., 'Invasion and Response: The Problems in Wales', in B. C.
Burnham and H. B. Johnson (eds.), Invasion and Response: The Case 01
Roman Britain (Oxford, 1979), 285-98.
Hopkins, K., Conquerors and Slaves: Sociological Studies in Roman His-
tory, i (Cambridge, 1978).
Horn, H. G., and Rüger, C. B. (eds.), Die Numider: Reiter und Könige
nördlich der Sahara (Cologne, 1979).
Hudson, J. c., 'Theory and Methodology in Comparative Frontier Stu-
dies', in D. H. Miller and J. 0. Steffen (eds.), The Frontier: Comparative
Studies (Norman, Okla., 1977), II-31.
Ilevbare, J. A., 'Family and Wornen in North Africa from the Fifth to the
First Century B.C.', Nigeria and Ihe Classics, 10 (1967-8), 26-44.
- - 'Some Aspects of Social Change in North Africa in Punic and
Roman Times', Museum Afrieum, 2 (1973), 24-44.
Isaac, B., 'The Meaning of "Limes" and "Limitanei" in Ancient Sourees',
JRS 78 (1988), 125-47.
- - The Limits 01 Empire: The Roman Army in the East, rev. edn.
(Oxford, 1992).
lanon, M., 'Recherches it Lambese', Ant. afi'. 7 (1973), 193-254, and 21
(19 85), 35-102.
- - 'Paysans et soldats', in C. M. WeHs (ed.), L'Alrique romaine: Les
Conlerences Vanier 1980. Roman Alrica: The Vanier Lectures 1980
(Ottawa, 1982), 51-68.
- - 'Remarques sur la frontiere de Numidie', in V. A. Maxfield and M. J.
Dobson (eds.), Roman Frontier Studies 1989." Proceedings 01 the XVth
International Congress 01 Roman Frontier Studies (Exeter, 199 1),482-4.
lohnson, D., The Roman Law of Trusts (Oxford, 1988).
Jones, G. D. B., "'Becoming Different Without Knowing It": The Role
and Development of Vici', in T. F. C. Blagg and A. C. King (eds.),
Military and Civilian in Roman Britain: Cultural Relationships in a
Frontier Provinee (Oxford, 1984), 75-91.
- - 'The UNESCO Libyan Valleys Survey: The Development of Settle-
ment Survey', in D. 1. Buck and D. 1. Mattingly (eds.), Town and
Country in Roman Tripolitania: Papers in Honour 01 Olwen Hackett
(Oxford, 1985), 263-89.
- - 'Concluding Remarks', in D. 1. Buck and D. 1. Mattingly (eds.), Town
and Country in Roman Tripolitania: Papers in Honour olOlwen Hackett
(Oxford, 1985), 307-1 I.
 266 Bibliography
Jones, G. D. B., and Barker, G. W W, 'Libyan Valleys Survey', Lib. Stud. I I
(1979-80), II-36.
Jones, M., 'Crop Production in Roman Britain', in D. Miles (ed.), The
RomanowBritish Countryside: Studies in Rural Settlement and Economy
(Oxford, 1982), i. 97-107.
- - 'Agriculture in Roman Britain: The Dynamics of Change', in M.
Todd (ed.), Research on Roman Britain 1960-89 (London, 1989), 12 7-34.
Jones, R. F. 1., 'A False Start? The Roman Urbanization of Western
Europe', WAreh. '9 (1987), 47-57.
- - 'Burial Customs of Rome and the Provinces', in 1. S. Wacher (ed.),
The Roman World (London and New York, 1987), ii. 812-31.
Kaddache, M., L'Algerie dans l'antiquite, 2nd edn. (Algiers, I982).
Kajanto, 1., The Latin Cognomina (Helsinki, 1965).
Kehoe, D. P., The Economics 0/ Agriculture on Roman Imperial Estates in
North Africa (Göttingen, 1988).
Kellner, H.-I, 'Die Möglichkeit von Rückschlüssen aus der Fundstatistik',
in W. Eck and H. Wolff (eds.), Heer und Integrationspolitik: Die
römischen Militärdiplome als historische Quelle (Cologne and Vienna,
I986), 24 I-8.
Keppie, L., 'Beyond the Northern Frontier: Roman and Native in Scot.
land', in M. Todd (ed.), Research on Roman Britain 1960-89 (London,
I989), 61-73-
- - 'Armies on Frontiers: Myth and Realities', in V. A. Maxfield and M. J.
Dobson (eds.), Roman Frontier Studies 198W Proceedings of the XVth
International Congress of Roman Frontier Studies (Exeter, 1991),455-7.
King, A. c., 'The Emergence of Romano·Celtic Religion', in T. F. C. Blagg
and M. Millett (eds.), The Early Roman Empire in the West (Oxford,
1990), 220-41.
Kotula, T., 'Cutte provincial et romanisation: Le Cas des deux Maureta.
nies', Eos, 63 (1975), 389-4°7.
- - 'Les Africains et la domination de Rome', Annales litteraires de
/'Universite de Besan,on, I86 (1976), 337-52.
Kraft, K., Zur Rekrutierung der Alen und Kohorten an Rhein und Donau
(Berne, 1951).
- - 'Zum Bürgerrecht der Soldatenkinder', Historia, IO (1961), 120-6.
Kuper, R., and Gabriel, B., 'Zur Urgeschichte des Maghreb', in H. G.
Horn and C. B. Rüger (eds.), Die Numider: Reiter und Könige nördlich
der Sahara (Cologne, 1979), 23-42.
Kurehin, B., 'Romans and Britons on the Northern Frontier: A Theoretical
Evaluation of the Archaeology of Resistance', in P. Rush (ed.), Theore.
tical Roman Archaeology: Second Conference Proceedings (Aldershot
and Brookfield, Vt., 1995), I24-31.
Lancel, S., 'Suburbures et Nicives: Une inscription de Tigisis', Libyca, 3
(1955), 28 9-98.
Bibliography
Laporte, I-P., Rapidum: Le Camp de la cohorte des Sardes en Maurhanie
Cesarienne (Sassari, 1989).
Larsen, C. S., 'Regional Perspectives on Mortuary Analysis', in L. A. Beck
(ed.), Regional Approaches to Mortuary Analysis (New York and
London, 1995), 247-64.
Lassere, I-M., 'Recherehes sur la chronologie des epitaphes pa'iennes de
l'Africa', Ant. afr. 7 (1973), 7-152.
- - Ubique populus: Peuplement et mouvements de population dans l'Afri-
que romaine de la chute de Carthage ci la./in de la dynastie des Severes
(14 6 av. C-235 p. C) (Paris, 1977).
- - 'Remarques onomastiques sur la liste militaire de Vezereos' in W S
Hanson and L. I F. Keppie (eds.), Roman Frontier Studies 1;7;
(Oxford, I980), 955-75.
- - 'Le Recrutement romain et les Musulames', in C. Lepelley (ed.),
Actes du IVe colloque international sur l'histoire et l'archeologie de
/'Ajrique du Nord (Strasbourg, 5-9 avril, 1988) (Paris, I991), 299-3 11.
- - 'Un conflit "routier": Observations sur les causes de la guerre de
Tacfarinas', Ant. afr. 18 (1992),11-25.
Lattimore, 0., Stu4ies in Frontier History (London and New York, 1962).
Lawless, R. 1., 'L'Evolution du peuplement, de l'habitat et des paysages
agraires du Maghreb', Annales de geographie, 81 (1972), 45 1- 64.
Le Bohec, Y, L'Armee romaine sous le Haut·Empire (Paris, 19 89).
- - La Troisieme Legion Auguste (Paris, 19 89).
- - Les Unites auxiliaires de l'armee romaine en Afrique Proconsulaire et
Numidie sous le Haut-Empire (Paris, 19 89).
Le Cceur, c., 'Les "Mapalia" numides et leur survivance au Sahara'
Hespiris, 24 (1937), 29-45. '
Leeuw, S. E. van der, 'Acculturation as Information Processing', in R.
Brandt and I Slofstra (eds.), Roman and Native in the Low Countries:
Spheres of Interaetion (Oxford, I983), 11-41.
Le Glay, M., Saturne africain: Histoire (Paris, 1966).
Leschi, L., Etudes d'epigraphie, d'archeologie et d'histoire africaines (Paris,
1957)·
Leveau, P., 'Une vallee agricole des Nemenchas dans l'Antiquite romaine:
L'Oued Hallail entre Djeurf et Aln Mdila', BCTH (r974-5), 10 3-2 1.
- - 'Paysans maures et villes romaines en Mauretanie cesarienne cen.
trale (Ja resistance des populations indigenes a la romanisation dans
l'arriere-pays ~e Caesarea de Mauretanie)" Melanges d'archeologie et
d'histoire de /'Ecole Franr:aise de Rome, 87 (1975), 857-71.
- - 'L'Opposition de la montagne et de la plaine dans l'historiographie
de l'Afrique du Nord antique', Annales de geographie, 86 (1977), 201-5.
- - 'La Situation coloniale de l'Afrique rornaine', Ann. ESC 33 (197 8),
89-9 2.
268 Bibliography Bibliography
Leveau, P., 'Occupation du sol, geosystemes et systemes sociau~: R?me et - - Legionmy Recruitment and Veteran Settlement during the Principate
ses ennemis des montagnes et du desert dans le Maghreb antlque , Ann. (London, 1983).
ESC 41 (1986), 1345-58. " - - 'Epigraphic Consciousness', JRS 75 (1985), 204-6.
-- 'Le Pastoralisme dans l'Afrique antique', In C. R. Whittaker (ed.), - - 'A Note on Conubium', in WEck and H. Wolff (eds.), Heer und
Pastoral Economies in Classical Antiquity (Cambridge, 1988), 177-95- Integrationspolitik: Die römischen Militärdiplome als historische Quelle
Levi della Vida, G., 'Sulle iscrizioni latino-libiche della Tripolitania', (Cologne and Vienna, 1986), 187-9.
Oriens antiquus, 2 (1963), 65-94. Manton, E. L., Roman North Africa (London, 1988 ).
- - 'Le iscrizioni neopuniche di Wadi el-Amud', Libya antiqua, I (1964), Marcone, A., 'Note sulla sedentarizzazione forzata delle tribit nomadi in
57-63. . Africa aHa luee di akune iscrizioni', in A. Mastino (ed.), L'A/rica
Lewis, D. R., Neither WalfNor Dog: American Indians, Envlronment, and Romana: AUi dei IX convegno di studio, Sassari (Sassari, 1992), 10 4-14.
Agrarian Change (New York and Oxford, 1994). . Marichal, R., 'Les Ostraca de Bu Njem', CRAI (I979), 436-5 2 .
Lieb H 'Die constitutiones für die stadtrömischen Truppen', In WEck - - Les Ostraca de Ru Njem (Tripoli, 199 2).
a;d H. Wolff (eds.), Heer und Integrationspolitik: Die römischen Mili- Marion, J., 'Les Ruines anciennes de la region d'Oujda', Bulletin d'arche-
tärdiplome als historische Quelle (Cologne an? Vienn~, ~986), 322-46: ologie marocaine, 2 (1957), II 7-73.
Lorcin, P. M. E., Imperial Identities: Stereotypmg, PrejUdlCe and Race Ln - - 'La Population de Volubilis a l'epoque romaine', Bulletin d'archeo-
Colonial Algeria (London and New York, 1995). .. . .. , lagie marocaine, 4 (1960), 133-87.
Luraschi, G., 'Sulla date e sui destinatari della lex MllllCla de hbens , Marshall, A. 1., 'Roman Women and the Provinees', Ancient SOciety, 6
Studia et Documenta Historiae et Iuris, 42 (1976), 321-70. (1975), I09-27.
Luttwak, E. N., The Grand Strategy oi the Roman Empire irom the First Mattingly, D. 1., 'New Perspectives on the Agricultural DeveIopment of
Century A.D. to the Third (Baltimore, 1976). Gebel and Pre-Desert in Roman Tripolitania', ROMM 41-2 (19 86),
McBurney, C. B. M., 'Libyan Role in Prehistory', in F. F. Gadallah (ed.), 46- 65.
Libya in History (Benghazi, 1971), 1-29. . . , . - - 'Libyans and the Limes: Culture and Soeiety in Roman Tripolitania',
MacKendrick, P., 'Roman Colonization and the FrontIer HypothesIs ,.lll Ant. aJr. 23 (1987), 71-94.
W D. Wyman and C. B. Kroeber (eds.), The Frontier in Perspectlve - - Tripolitania (Ann Arbor, 1994).
(Madison, 1957), 3-19. . - - 'First Fruit? The Olive in the Roman World', in G. Shipley and J.
- - The North AJrican Stones Speak (Chapel H!ll, NC, 1980) . . Salmon (eds.), Human Landscapes in Classical Antiquity: Environment
Mackie, N., 'Urban Munificence and the Growth ofUrban ConsclOusness and Culture (London and New York, 1996), 21 3-53.
in Roman Spain', in T. F. C. Blagg and M. Millett (eds.), The Early - - 'From One Imperialisrn to Another: Imperialism and the Maghreb',
Roman Empire in the West (Oxford, 1990), 179-92. in 1. Webster and N. Cooper (eds.), Roman Imperialism: Post-Colonial
MacMullen, R., Soldier and Civilian in the Later Roman Empire Perspectives (Leicester, 1996), 49-69.
(Cambridge, Mass., 1963). - - and Hitchner, R. B., 'Roman Africa: An Archaeologieal Review',
- - 'Rural Romanization', Phoenix, 22 (1968), 337-41. JRS 85 (1995), 165-213.
- 'Market-Days in the Roman Empire', Phoenix, 24 (1970), 333-41. - - and Jones, G. D. B., 'A New Clausura in Western Tripolitania: Wadi
- 'The Epigraphic Habit in the Roman Empire', AJP 103 (1982), Skiffa South', Lib. Stud. 17 (1986), 87-9 6.
233-4 6. Maxfield, V A., 'Systems ofReward in Relation to Military Diplomas', in
- - Changes in the Roman Empire: Essays in the Ordinary (Princeton, WEck and H. Wolff(eds.), Heer un,d Integrationspolitik: Die römischen
199 0). Militärdiplome als historische Quelle (Cologne and Vienna, 1986 ),26-43.
Mahboubi, M., 'Les Elites municipales de la Numidie; deux groupes: - - 'Conquest and Aftermath', in M. Todd (ed.), Research on Roman
etrangers a la eite et veterans', ANRW 2/10/2 (1982), 67~-81. , . Eritain 1960-89 (London, 1989), 19-29.
Mann, 1. c., 'The Development of Auxiliary and Fleet Dlplqmas , EPI- - - 'The Frontiers of the Roman Empire: Some Recent Work', JRA 2
graphische Studien, 9 (I972), 233-41. (I9 89), 334-46.
- - 'The Frontiers of the Principate', ANRW 2/1 (1974), 508-33. - - and Dobson, M. J. (eds.), Roman Frontier Studies 1989: Proceedings
- 'Power, Force and the Frontiers ofthe Empire', JRS 69 (1979),175-83. oi the XVth International Congress oi Roman Frontier Studies (Exeter,
199 1).
 Bibliography
Bibliography
M'Charek, A., Aspec,ts de l' evolution demographique et sodale a Mactaris
aux I/e et IIIe siedes apo J C. (Tunis, 1982). - - The Romanization of Britain: An Essay in Archaeological Interpreta~
Merlin, A. (ed.), Inscriptions latines de la Tunisie (Paris, 1944). . tion (Cambridge, 1990).
Mertens, 1., 'The Military Origins of Same Roman Settlements m - - 'Romanization: Historical Issues and ArchaeologicaI Interpreta-
Belgium', in B. Hartley and 1. S. Wacher (eds.), Rame and her ,!orth~rn tions', in T. F. C. Blagg and M. Millett (eds.), The Early Roman Empire
Provinces: Papers Presented to Sheppard Frere in Honour o[ hls !--etl~e­ in the West (Oxford, 1990), 35-41.
ment [rom the Chair 0/ the Archaeology 0/ the Roman Empire, Umverslly Mirkovic, M., 'Die römische Soldatenehe und der "SoIdatenstand"',
of Oxford, 1983 (Gloueester, 1983), 155-65, . Zeitschrift für Papyrologie und Epigraphik, 40 (1980), 259-71.
Meyer, E., 'Explaining the Epigraphic Habit in the Roman EmpIre: The - - 'Die Entwicklung und Bedeutung der Verleihung des Conubium', in W
Evidenee of Epitaphs', JRS 80 (1990), 74-96. Eck and H. Wolff (eds.), Heer und Integrationspolitik: Die römischen
Middleton, P, 'Army Supply in Roman Gaul: An Hypothesis for Roman Militärdiplome als historische Quelle (Cologne and Vienna, 1986), 16 7-86.
Britain', in B. C. Burnham and H. B. Johnson (eds.), Invasion and MitchelI, R. D., 'Introduction: Revisionism and Regionalism', in R. D.
Response: The Case of Roman Britain (Oxford, 1979),81-97. Mitchell (ed.), Appalachian Frontiers: Settlement, Society and Develop~
- - 'The Roman Army and Long~Distance Trade', in P. Garnsey and C. ment in the Preindustrial Era (Lexington, Mass., 1991), 1-22.
R. Whittaker (eds.), Trade and Famine in Classical Antiquity Mitteis, L., and Wilcken, U., Grundzüge und Chrestomathie der Papyrus~
(Cambridge, 1983), 75-83. . kunde, 4 vols. (Leipzig, 1912).
Mikesell, M., 'Comparative Studies in Frontier History', 1ll R. H?fstadter Mocsy, A., 'Das territorium legionis und die canabae in Pannonien', Acta
and M. Lipset (eds.), Turner and the Sociology of the Frontier (New Archaeologica Academiae Scientiarum Hungaricae, 3 (1953), 179-99.
York, 1968), 152-71. - - Gesellschaft und Romanisation in der römischen Provinz Moesia Super-
Mikl Curk, L, 'Natives, Romans and Newcomers in th~ Eastern .Alp,s ior (Amsterdam, 1970).
during the 2nd Century (The Role of the Army in Ethmc InteractIon) , - - 'Das Namensverbot des Kaisers Claudius (SueL Claud. 25, 3)', Klio,
in V A. Maxfield and M. J. Dobson (eds.), Roman Frontier Studies 52 (1970), 287-94-
[9 89." Proceedings ofthe XVth International Congress of Roman Frontier Mori, F., 'Prehistoric Saharan Art and Cultures in the Light of Discov-
Studies (Exeter, 1991),248-51. . eries in the Aeacus Massif (Libyan Sahara)', in F. F. Gadallah (ed.),
Milburn, M., 'Romans and Garamantes: An Enquiry into.Contacts', l~ D. Libyo in History (Benghazi, 1971), 3 1-9.
1. Buck and D. J. Mattingly (eds.), Town and Country m Roman Tnpo- Morizot, P., 'Inscriptions inedites de rAures', Zeitschrift für Papyrologie
litania: Papers in Honour of Olwen Hackelt (Oxford, 1985),241-61. und Epigraphik, 22 (1976), 13 8-68.
Miles, D., 'The Romano-British Countryside', in M. Todd (ed.), Research - - 'Vues nouvelles sur rAures antique', CRAI (1979),3°9-37.
on Roman Britain [960-89 (London, 1989), 115-26 . . - - 'Les Inscriptions de Tazembout (Aures): Apen;u sur un village
Millar, F., 'The Date of the Constitutio Antoniniana', Journal of Egyptwn romain de haute montagne au lIIe siede', BCTH (19 84-5),69-100.
Archaeology, 48 (1962), 124-31. " . - - 'Economie et societe en Numidie meridionale: L'Exemple de I'Aures',
- - 'Loeal Cultures in the Roman Empire: Libyan, PUllIe and Lahn 1ll in A. Mastino (ed.), L'Africa Romana: Attf dei VIII convegno di studio,
Roman Afriea', JRS 58 (1968), 126-34. Cagliari (Sassari, 1991), 429-4 6 .
- - Review of H. Wolff, Die Constitutio Antoniniana und Papyrus - - 'Le Reseau de communications de Ia lIIe legion de Lambese au
Gissensis 40 1 (Cologne, 1976), JRS 67 (1977), 235. Sahara a travers l'Aures', in C. Lepelley (ed.), Actes du IVe colloque
--- 'Emperors, Frontiers and Foreign Relations, 31 S.C.-A.D. 378', international sur l'histoire et l'archeologie de I'Afrique du Nord (Stras~
Britannia, 13 (1982), 1-23. . bourg, 5-9 avril. 1988) (Paris, 199 1),4°9-26.
MilIer, D. H., and Savage, W W, 'Ethnic Ster~otypes an~ the ,Fr.ontler: A - - 'L'Aures et I'olivier', An!. ufr. 29 (1993),177-240 .
Comparative Study of Roman and Amencan Expenence: In D. ~. Morris, 1., Death~Ritual and Social Structure in Classical Antiquity (Cam~
Miller and J. 0. Steffen (eds.), The Frontier: Comparatlve Studles bridge, 1992).
(Norman, Okla., 1977), 109-37. . Morris, 1., 'Changing Fashions in Roman Nomenclature in the EarIy
Millett M. 'Forts and the Origins of Towns: Cause or Effeet?', 1ll T. F. C. Empire', Listy Filologicke, II (1963), 34-4 6.
Blag~ a~d A. C. King (eds.), Military and Civilian in Roman Britain: Musset, L., The Germanie Invasions: The Making of Europe, AD 400-600,
Cultural Relationships in a Frontier Province (Oxford, 1984), 65-74. trans. E. and C. James (University Park, Pa., 1975).
 272 Bibliography
Nash-Williams, V. E., The Roman Frontier in Wales (Cardiff, 1954).
Nesselhauf, H., 'Das Bürgerrecht der Soldatenkinder', Historia, 8 (I959),
434-4 2.
Okun, M. L., The Early Roman Frontier in the Upper Rhine Area: Assim-
ilation and Acculturation on a Roman Frontier (Oxford, I9 89).
- - 'An Example of the Process of Acculturation in the Early Roman
Frontier', Oxford Journal oj Archaeology, 8/1 (1989),41-54.
-- 'Pluralisrn in Germania Superior', in V. A. Maxfield and M. J.
Dobson (eds.), Roman Frontier Studies 1989: Proceedings 0/ the XVth
International Congress o[ Roman Frontier Studies (Exeter, 199 1),435-8.
0rsted, P, Roman Imperial Economy and Romanization (Copenhagen,
1985).
Pavis d'Escurac, H., 'Irrigation et vie paysanne dans l'Afrique du Nord
antique', Ktema, 5 (1980), 177-91.
- - 'Nundinae et vie rurale dans l'Afrique romaine', BCTH (1981),251-9.
Petrikovits, H. von, 'Lixae', in W S. Hanson and L. J. F. Keppie (eds.),
Roman Frontier Studies /979 (Oxford, 1980), 1027-34.
Pflaum, H.wG., 'Remarques sur l'onomastique de Castellum Celtianum',
in E. Swoboda et al. (eds.), Carnuntina: Ergebnisse der Forschung über
die Grenzprovinzen des römischen Reiches: Vorträge beim internationalen
Kongress der Altertumsforscher Carnuntum /955 (Graz and Cologne,
1956), 126-51; also in Pflaum, L'Afrique romaine, 87-II2.
- -- 'Onomastique de Cirta', in R. LaurwBelart (ed.), Limes-Studien:
Vorträge des 3 internationalen Limeskongresses in Rhein/eiden/Basel
(Basel, 1959),96-133; also in Pflaum, L'Ajrique romaine, 161-98.
-- 'La Romanisation de l'Afrique', Vestigia, I7 (1973), 55-72; also in
Pflaum, L'Afrique romaine, 375-92.
-- 'Remarques sur l'onomastique de Castellum Tidditanorum', BCTH
(1974-5),9-44·
- - L'Afrique romaine: .ktudes epigraphiques (Paris, 1978).
Pieard, G.-C., Castellum Dimmidi (Paris, 1957).
Potter, D., 'Empty Areas and Roman Frontier Poliey', AJP 113 (1992),
26 9-74.
Prescott, J., Boundaries and Frontiers (London, 1978).
Purcell, N., 'The Creation of Provincial Landscape: The Roman Impact
on Cisalpine Gaul', in T. F. C. Blagg and M. Millett (eds.), The Early
Roman Empire in the West (Oxford, 1990),7-29.
Rachet, M., Rome et les Berberes: Un probleme militaire d'Auguste cl
Dioclhien (Brussels, 1970).
Rakob, F., 'Numidische Königsarchitektur in, Nordafrika', in H. G. Horn
and C. B. Rüger (eds.), Die Numider: Reiter und Könige nördlich der
Sahara (Cologne, 1979), II 9-7 I.
Raven, S., Rome in Africa, 3rd edn. (London and New York, 1993).
Bibliography
Rawson, B., 'The Roman Family', in B. Rawson (ed.), The Family in
Ancient Rome: New Perspectives (Ithaca, NY, 1986), I-57.
Rebuffat, R., 'Graffiti en Libyque de Bu Njem', Libya antiqua, 11-12
(1975), 165-87.
- - 'Une zone militaire et sa vie economique: Le Limes de Tripolitaine',
in Armees et fiscalite dans le monde antique (Paris, 1977), 395-4 19.
- - 'La Frontiere romaine en Afrique: Tripolitaine et Tingitane', Ktema,
4 (1979), 225-47.
- - 'Au-delil des camps romains d'Afrique mineure: Renseignement,
contrale, penetration', ANRW vrolz (1982), 474-5 13.
Redmer, H., 'Naturlandschaft und Kulturlandschaftswandel Nordafrikas'
in H. G. Horn and C B. Rüger (eds.), Die Numida' Reiter und König;
nördlich der Sahara (Cologne, 1979), 5-21.
Reece, R., 'Romanization: A Point of View', in T. F. C. Blagg and M.
Millett (eds.), The Early Roman Empire in the West (Oxford, 199 0 ),3 0 -4.
Reynolds,1. M., 'Inscriptions in the PrewDesert of Tripolitania', in D. J.
Buck and D. J. Mattingly (eds.), Town and Country in Roman Tripoli-
tania: Papers in Ronour of Olwen Hackett (Oxford, 1985), 23-5.
Riccobono, S., et al. (eds.), Fontes luris Romani Anteiustiniani, 2nd edn.
(Florenee, 1940-3).
Romanelli, P., Storia delle province romane dell'Africa (Rome, 1959).
Rössler, 0., 'Die Numider-Herkunft, Schrift, Sprache', in H. G. Horn
and C. B. Rüger (eds.), Die Numider: Reiter und Könige nördlich der
Sahara (Cologne, 1979), 89-97.
Roxan, M., Roman Military Diplomas 1954-1977 (London, 1978).
- - 'The Distribution of Roman Military Diplomas', Epigraphische Stu-
dien, 12 (1981), 265-86.
- - Roman Military Diplomas 1978-1984 (London, 19 85).
- - 'Observations on the Reasons for the Changes in Formula in
Diplomas circa AD 140', in WEck and H. Wolff (eds.), Heer und
Integrationspolitik: Die römischen Militärdiplome als historische Quelle
(Cologne and Vienna, 1986),265-92.
- - 'Women on the Frontiers', in V A. Maxfield and M. J. Dobson (eds.),
Roman Frontier Studies 1989: Proceedings of the XVth International
Congress of Roman Frontier Studies (Exeter, 199 1 ),4 62-7.
Roymans, N., 'The North Belgic Tribes in the Ist Century B.C: A Histor-
ieal-Anthropologieal Approach', in R. Brandt and 1. Slofstra (eds.),
Roman and Native in the Low Countries: Spheres o/lnteraction (Oxford,
19 83), 43-69.
Ruprechtsberger, E. M., Die Garamanten (Mainz am Rhein, 1997).
Saddington, D. B., 'The Parameters of Romanization', in V A. Maxfie1d
and M. 1. Dobson (eds.), Roman Frontier Studies 1989: Proceedings of
the XVth International Congress o[ Roman Frontier Studies (Exeter,
199 1),4 13-18.
 274 Bibliography
Salama, P, -Les Voies romaines de I'Ajrique du Nord (Algiers, 1951).
- - 'Les Deplacements successifs du limes en Mauretanie Cesarienne' , in 1.
Fitz (ed.), Akten des XI internationalen Limeskongresses, SzekesJehervar,
/976 (Budapest, 1977),577-95.
- - 'Quelques incursions dans la zone occidentale du limes de Numidie',
Ant. aJr. 27 (1991), 93-I05.
Saller, R. P., Patriarchy, Property and Death in the Roman Family
(Cambridge, 1994).
- - and Shaw, B. D., 'Tombstones and Roman Farnily Relations in the
Principate: Civilians, Soldiers and Slaves', JRS 74 (1984), 124-56.
Salway, P., The Frontier People 0/ Roman Britain (Cambridge, 19 65).
Sasse, c., Die Constitutio Antoniniana (Wiesbaden, 195 8).
Seidl, E., Rechtsgeschichte Ägyptens als römische Provinz (Sankt Augustin,
1973)·
Shaw, B. D., 'Climate, Environment and Prehistory in the Sahara', W
Arch. 8 (1976), 133-49; also in Shaw; Environment and Society.
- 'Rural Periodic Markets in Roman North Africa as Mechanisms of
Social Integration and Control', Research in Economic Anthropology, 2
(I979),9 I- I I 7· . .
- - 'The Camel in Roman North Africa and the Sahara: HIstory, BlOI-
ogy and the Human Economy', Bulletin de I'Institut Fondamental d'Afri-
que Noire, 41 (1979), 663-721; also in Shaw, Environmen~ and Soc~ety.
- - 'Archaeology and Knowledge: The History of the Afncan Provlllces
of the Roman Empire', Florilegium, 2 (1980), 28-60; also in Shaw,
Environment and Society.
- - 'Climate, Environment and History: The Case of Roman North
Africa', in T. M. L. Wigley, M. 1. Ingram, and G. Farmer (eds.), Climate
and History: Studies in Past Climates and their Impact on Man
(Cambridge, 1981), 379-403; also in Shaw, Environment and Society.
- .. 'The EIder Pliny's African Geography', Historia, 30 (1981), 424-7 1;
also in Shaw, Rulers, Nomads and Christians.
- - 'Rural Markets in North Africa and the Political Economy of the
Roman Empire', Ant. afr. 17 (1981), 37-83; also in Shaw, Rufers,
Nomads and Christians.
- - 'Fear and Loathing: The Nomad Menace and North Africa', in C. M.
Wells (ed.), L'Afrique romaine: Les Conferences Vanier 1980. Roman
Africa: The Vanier Lectures 1980 (Ottawa, 1982), 29-50; also in Shaw,
Rulers, Nomads and Christians.
- - 'Lamasba: An Ancient Irrigation Community', Ant. afr. 18 (19 82 ),
61-103; also in Shaw, Environment and Society.
- - 'Eaters of Flesh, Drinkers of Milk: The Ancient Mediterranean
Ideology of the Pastoral Nomad', Ancient SOciety, 13-14 (19 82 -3),
5-3 1; also in Shaw, Rufers, Nomads and Christians.
Bibliography 275
- - 'Soldiers and Society: The Army in Numidia', Opus, 2/1 (19 83),
133-60; also in Shaw, Rufers, Nomads and Christians.
- - 'Water and Society in the Ancient Maghrib: Technology, Property
and Development', Ant. afr. 20 (1984), 121-73; also in Shaw, Environ-
ment and Society.
- - 'Autonomy and Tribute: Mountain and Plain in Mauretania Tingi-
tana', RaMM 41-2 (1986), 66-89; also in Shaw, Ru!ers, Nomads and
Christians.
- - 'The Noblest Monuments and the Smallest Things: WeHs, Walls and
Aqueducts in the Making of Roman Africa', in A. T. Hodge (ed.),
Future Currents in Aqueduct Studies (Leeds, 1991), 63-91; also in
Shaw, Environment and Society.
- - 'The Structure of Local Society in the Early Maghrib: The Elders',
The Maghrib Review, 16 (1991), 18-54; also in Shaw, Rufers, Nomads
and Christians.
--- Environment and Society in Roman North Africa: Studies in History
and Archaeo!ogy (Brookfield, Vt., 1995).
- - Rulers, Nomads and Christians in Roman North Alrica (BrookfieId,
Vt., 1995).
Sheldon, R., 'Romanizzazione, acculturazione e resistenza: Problemi con-
cettuali nella storia deI Nordafrica', Diafoghi di Archeofogia, NS 4
(1982), I02-6.
Sherwin-White, A. N., 'Geographical Factors in Roman Aigeria', JRS 34
(r944), I-IO.
--- The Roman Citizenship, 2nd edn. (Oxford, 1973).
- - 'The Tabula of Banasa and the Constitutio Antoniniana', JRS 63
(1973), 86-98.
Slofstra, 1., 'An Anthropological Approach to the Study of Romanization
Processes', in R. Brandt and 1. Slofstra (eds.), Roman and Native in the
Low Countries: Spheres olInteraction (Oxford, 1983),7 1- 104.
Snape, M. E., 'Roman and Native: Vici on the North British Frontier', in
V. A. MaxfieId and M. 1. Dobson (eds.), Roman Frontier Studies /9 8g:
Proceedings of the XVth International Congress 01 Roman Frontier Stu-
dies (Exeter, 1991), 468-7I.
Sommer, C. S., The Military Vici in Roman Britain (Oxford, 19 84).
Soyer, 1., 'Les Centuriations romaines en Algerie orientale', Ant. afr. 10
(1976), I07-80.
Speidei, M., 'Africa and Rome: Continuous Resistance?', Proceedings of
the Alrican Cfassicaf Association, 13 (1975), 36-8.
- - 'The Soldiers' Homes', in WEck and H. Wolff (eds.), Heer und
Integrationspofitik: Die römischen Militärdipfome afs historische Quelle
(Cologne and Vienna, 1986), 467-8I.
- - 'The Roman Army in North Africa', JRA 5 (1992), 401 -7.
 Bibliography
Stallard; P. Y., Glittering Misery: Dependents o[ the Indian Fighting Army
(San Rafael, Calif., 1978). . - - 'L'Idee de frontiere au Sahara d'apres les donnees archeologiques',
Steffen, 1. 0., Comparative Frontiers: A Proposal lor Studymg the
American West (Norman, Okla., 1980).
Struck, M., 'Analysis of Sodal and Cultural Diversity on Rural Burial Sites
in North-Eastern Raetia', in P. Rush (ed.), Theoretical Roman Archae-
ology: Second Conference Proceedings (Aldershot and Brookfield, Vt.,
1995), 70-80.
Swanson,1. T., 'The Myth ofTrans-Saharan Trade during the Roman Era',
International Journal 0/ African Historical Studies, 8 (1975), 583- 600.
Syme, R., 'Tacfarinas, the Musulamii and Thubursicu', in P. Coleman-
Norton (ed.), Studies in Roman Social and Economic History in Hanor 0/
A. C. Johnson (Princeton, 1951), 113-30; also in R. Syme, Roman
Papers, ed. E. Badian and A. R. Birley (Oxford, 1979-91), i. 218-30.
- - Colonial EWes: Rome, Spain and the Americas (Landon, 195 8).
Taubenschlag, R., 'Citizens and Non-Citizens in the Papyri', in Scritti di
diritto romano in onore di C. Ferrini (Milan, 1948), iii. 166-73.
- - The Law 0/ Greco-Roman Egypt in the Light 0/ the Papyri, 2~d edn.
(Warsaw, 1955).
Teutsch, L., 'Gab es Doppelgemeinden im römischen Afrika?', Revue
internationale des droUs de l'antiquite, 8 (1961), 281-356.
- - Das Städtewesen in Norda/rika in der Zeit von C Gracchus bis zum
Tode des Kaisers Augustus (Berlin, 1962).
Thebert, Y., 'Romanisation et deromanisation en Afrique: Histoire deco-
lonisee ou histoire inversee?', Ann. ESC 33 (1978), 64-82.
Thompson, L., and Lamar, H., 'Comparative Frontier History', in ~.
Lamar and L. Thompson (eds.), The Frontier in History: North Amenca
and Southern Africa Compared (New Haven and London, 1981), 3-1~.
Thampson, L. A., 'Some Observations on Personal Nomenclature In
Roman Africa', Nigeria and the Classics, 10 (1967-8), 45-58. . (195 2),401-22.
- - 'Roman and Native in the Tripolitanian Cities in the Early EmpIre',
in F. F. Gadallah (ed.), Libya in History (Benghazi, 1971), 235-50.
Todd, M., 'The Early Cities', in M. Todd (ed.), Research on Roman Britain
19 60- 89 (London, 1989), 75-89.
Treggiari, S., Roman Marriage: lusti Coniuges /rom {he Time 0/ Cicero to
the Time of Ulpian (Oxford, 1991). . . ,
Trousset, P, Recherehes sur le limes Tripolitanus du Chott el-Djend a la
/rontiere tuniso-libyenne (Paris, 1974).
- - 'Signification d'une frontiere: Nomades et sedentaires dans la zone
du limes d'Afrique', in W S. Hanson and L. J. F Keppie (eds.), Roman
Frontier Studies 1979 (Oxford, 1980),931-42.
- . 'L'Image du nomade saharien dans l'historiographie antique', Pro-
duction pastorale et societe, 10 (1980), 97- 10 5.
Bibliography
in P.-R. Baduel (ed.), Enjeux sahariens: Tabte ronde du Centre de
Recherches et d'Etudes sur les Socihes Mediterraneennes (Paris, 1981),
47-78. .. d I' . ..,
- - 'La Franchissement des chotts du sud-TUlllslen ans antIqUlte,
Ant. afr. 18 (1982), 45-59.
- - 'Note sur un type d'ouvrage lineaire du limes d'Afrique', BCTH
(1984), 183-98.
- - 'Limes et frontiere climatique', in Actes du IIle colloque international
sur I'histoire et I'archeologie de I'Afrique du Nord (Montpelliel; 1-15 avril,
1985) (Paris, 1986),55-84.
- - 'De la montagne au desert: Limes et maitrise de l'eau', ROMM 41-2
(19 86 ),90- 1I 5·
- - 'Les Oasis presahariennes dans l'Antiquite: Partage de l'eau et divi-
sion du temps', Ant. afr. 22 (1986), 16 3-93.
Trow, S. D., 'By the Northern Shores of Ocean: Some Observations on
Acculturation Process at the Edge of the Roman World', in T. F C.
Blagg and M. Millett (eds.), The Early Roman Empire in the West
(Oxford, 1990), 103-18.
Van der Veen, M., 'Native Communities in the Frontier Zone: Uniformity
or Diversity?', in V. A. Maxfield and M. J. Dobson (eds.), Roman
Frontter Studies 1989: Proceedings 0/ the XVth International Congress
0/ Roman Frontier Studies (Exeter, 199 1),44 6-5 0 .
Vertet, H., 'Religion populaire et rapport au pouvoir d'apres les statuettes
d'argile sous l'empire romain', in A. Daubigney (ed.), Archeologie et
rapports sociaux en Gaule (Paris, 1984),7 2-122.
Visscher, F de, 'Conubium et Civitas', Jura, 2 (1951), 140 -4.
- - 'Conubium et Civitas', Revue internationale des droits de l' antiquite, I
.
Vittinghoff, F, 'Die rechtliche Stellung der canabae legionis und dIe
Herkunftsangabe castris', Chiron, I (1971), 299-3 1 8.
- - 'Militärdiplome, römische Bürgerrechts- und Integrationspolitik der
Hohen Kaiserzeit', in WEck and H. Wolff (eds.), Heer und Integra-
tionspolitik: Die römischen Militärdiplom,e als historische Quelle
(Cologne and Vienna, 1986), 535-55.
Volterra, E., 'La nozione giuridica de! conubium', in Studi in memoria di
E Albertario (Milan, 1953), ii. 347-84.
Ward-Perkins, J. B., 'Pre-Roman Elements in the Architecture of Roman
Tripolitania', in F. F. Gadallah (ed.), Libya in History (Benghazi, 1971),
101-16.
Watson, A., The Law 0/ Persons in the Later Roman Republic (Oxford,
1967).
Watson, G. R., The Roman Soldier (London, 19 69).

Bibliography
Waurick, G" 'Die Schutzwaffen im numidischen Grab von Es Soumä~', .in
H. G. Horn and C. B. Rüger (eds.), Die Numider: Reiter und Konrge
nördlich der Sahara (Cologne, '979), 30 5-3 2,
Weaver, P. R. c., 'Where Have All the Junian Latins Gone?: Nomenclature
and Status in the Early Empire', Chiron, 20 (1990), 275-3 0 5.
Webb, W P., The Great Plains (New York, 193 1).
Webster G. A. 'A Note on New Discoveries at Viroconium (Wroxeter)
Which May Have a Bearing on Hadrian's Frontier Policy in Britain', in W
S. Hanson and L. 1. F. Keppie (eds.), Roman Frontter Studies 1979
(Oxford, 1980), 29l-6.
Webster, J., 'Roman Imperialism and the Post-Imperial Age', in 1. Webster
and N. Cooper (eds.), Roman Imperialism: Post-Colonial Perspectives
(Leicester, '996), l-18. . .
__ 'Necessary Comparisons: A Post-Colonial Approach to RehglOus
Syncretism in !he Roman Provinces', W Arch. 28/3 (l997), 324-38.
Wells, C. M., The German Policy of Augustus (Oxford, 197 2).
__ 'The Impact of the Augustan Campaigns on Germany', in D. M.
Pippidi (ed.), Assimilation et resistance ala culture grec~-romaine da.ns le
monde ancien: Travaux de Vle Congres International d'Etudes Classzques
(Madrid, Sept. 1974) (Buchares! and Paris, 1976), 421 -31.
__ 'The Problems of Desert Frontiers' , in V. A. Maxfield and M. 1.
Dobson (eds.), Roman Frontier Studies 1989: Proceedings of the
XVth International Congress of Roman Frontier Studies (Exeter,
1991), 47 8- 8 1. . .
__ '''The Daughters ofthe Regiment": Sisters and Wlves m the Rom~n
Army', forthcoming in the proceedings of the XVIth Roman FrontIer
Studies Conference (August, 1995)·
Wheeler, E. L., 'Methodological Limits and the Mirage of Roman Strat-
egy', Journal of Military History, 57 (l993), 7-4l, 2l5-:40., . ,
Whitehorne, 1. E. G., 'Ovid, A.A. I. 101-32, and Soldlers Marrmges,
Liverpool Classical Monthly, 4 (l979), 157-8. . .
Whittaker, C. R., 'Agri Deserti', in M. I. Finley (ed.), Studles ln Roman
Property (Cambridge, 1976), l37-65, 193-200; also in Whittaker, Land,
City and Trade. . gerrechtspolitik der Kaiserzeit' , in Studien zur antiken Sozialgeschichte:
__ 'Land and Labour in North Africa', Klio, 60 (1978), 33l-62; also In
Whittaker, Land, City and Trade.
__ 'Rural Labour in Three Roman Provinces', in P. Garnsey (ed.), Non-
Slave Labour in the Greco-Roman World (Cambridge, 1980), 73-99; also
in Whittaker, Land, City and Trade.
__ 'Trade and Frontiers ofthe Roman_Empire', in P. Garnsey and C. R.
Whittaker (eds.), Trade and Famine in Classical Antiquity (Cambridge,
'983), lI0-27·
Bibliography 279
- - 'Introduction', in C. R. Whittaker (ed.), Pastoral Economies in Clas-
sieal Antiquity (Cambridge, 1988), 1-5.
- - 'Supplying the System: Frontiers and Beyond', in 1. C. Barrett, A. P.
Fitzpatrick, and L. Macinnes (eds.), Barbarians and Romans in North-
West Europe: From the Later Republic to Late Antiquity (Oxford, 1989),
64-80.
- - 'Do Theories of the Ancient City Matter?', in T. 1. Cornell and H. K.
Lomas (eds.), Urban Society in Roman Italy (London, 1993), 1-20; also
in Whittaker, Land, City and Trade.
- - Land, City ond Trade in the Roman Empire (Brookfield, VI., 1993).
- - Frontiers 0/ the Roman Empire: A Social and Economic Study (Balti-
more, 1994).
- - 'Integration of the Early Roman West: The Example of Africa', in 1.
Metzler et al. (eds.), Integration in the Early Roman West: The Role of
Culture and Ideology (Luxembourg, 1995), 19-32.
Wightman, E. M., 'Military Arrangements, Native Settlements and
Related Developmen!s in Early Roman Gaul', Helinium, 17 (l977),
105-26.
Wild,1. P., 'Roman and Native in Textile Technology', in B. C. Burnham
and H. B. Johnson (eds.), Invasion and Response: The Case 0/ Roman
Britain (Oxford, 1979), 123-31.
Willems, W. 1. H., 'Romans and Batavians: Regional Developrnents at the
Imperial Frontier', in R. Brandt and 1. Slofstra (eds.), Roman and Native
in the Low Countries: Spheres o/Interaction (Oxford, 1983), 105-28.
Williams, D., The Reach 0/ Rome: A History 0/ the Roman Imperial
Frontier Ist-5th Centuries AD (New York, 1996).
Wilson, R. 1. A., Review of M. Millett, The Romanization 0/ Britain: An
Essay in Archaeological Interpretation (Carnbridge, 1990), JRS 82
(l992),290-3.
Wolff, H., 'Zu den Bürgerrechtsverleihungen an Kinder von Auxiliaren
und Legionaren', Chiron, 4 (l974), 470-5l0.
- - Die Constitutio Antoniniana und Papyrus Gissensis 40 I (Cologne,
1976).
- - 'Zum Erkenntniswert von Namenstatistiken für die römische Bür-
Festschrift F Vittinghoff (Cologne, '980), 229-55.
Woolf, G., 'The Unity and Diversity of Romanisation', JRA 5 (r992),
349-5 2 .
- - 'The Formation of Roman Provincial Cultures', in 1. Metzler et al.
(eds.), Integration in the Early Roman West: The Role 0/ Culture and
Ideology (Luxembourg, 1995), 9-l8.
- - 'Monumental Writing and the Expansion of Roman Society in the
Early Empire', JRS 86 (l996), 22-39.
- - 'Beyond Romans and Natives', W Arch. 28/3 (l997), 339-50.
x Prefaee
The ongms of this book go back some fifteen years, in the
course of which I have contracted a great many debts, intellectual
and personaL Susan Treggiari and Colin Wells have taught me Contents
much of what I know about the Roman world, and much else
besides. They continue to inspire me. I am grateful also to Benjamin
Isaac and Greg Woolf, who reviewed the whole of the manuscript
and did much to improve it, and to Brent Shaw, who read an
List of Figures xm
earlier version of the chapter on marriage, and encouraged me to
think that I had something worth saying. List of Tab!es xiv
I gratefully acknowledge my indebtedness also to the College of
Abbreviations xv
Letters and Science at Montana State University, for several grants
in aid of research, to my colleagues in the Department of History I. Pre-Roman Algeria
and Philosophy, who have provided me with a wonderfully exhi- The Landscape 3
larating environment in which to work, to the editors of the Press, Economy and Society 9
whose help has been unstintingly generous, and to Phoenix, for
2. The Roman Frontier-Zone 24
allowing me to rework material which it published originally in
Locating the Roman Frontiers 28
1990. Particular thanks are owed to Tom WesseI, for advice about
Native American agricultural practices, to Warren Esty, for help The Development of the Frontier-Zone 35
with statistics, and to Richard Griggs, who prepared several of the The Purpose of the Frontier-System 58
ConcJusion: Roman Objectives in the Frontier-Zone 73
figures.
I shall always be indebted to my wife Jeanette, without whose 3· Measuring Romanization 75
unfailing support I should have despaired of finishing this book Sources and Methods 78
(and of much else). For her tireless patience in teaching me to Unworkable Models 82
recognize what really matters, in all that constitutes our everyday Intermarriage 99
world, I cannot expect to repay her.
4· Husbands and Wives in the Frontier-Zone 101
D.C. Sources and Methods 1°3
Marriage-Patterns 119
Law and Social Practice 134
5· Consequences of the Roman Occupation 14!
Agriculture 142
Trade 153
Society 155
6. ConcJusion: The Limits of Acculturation 15 8
Appendices
I. Marriage-Epitaphs of the Frontier-Zone
(e·50 BC-AD 250) 162
2. Marriage-Epitaphs of Thubursicu Numidarum
(e·50 BC-AD 250) 209
xii Contents
3. Marriage-Epitaphs that Record Roman(ized)
Wives and Husbands 220
4. Dated Epitaphs of the Frentier-Zone 225 List of Figures
Bibliography 0/ Modern Works Cited in the Notes 254
Index 0/ Passages 281
Index 0/ Proper Names 282 LI. Physical geography of Algeria 6
Index 0/ Sub}ects 1.2. Bio-climatic zones of Algeria 7
1.3· Algerian transhumant reutes 14
1.4· The distribution of indigenous tribes in ancient
Algeria 19
2.1. The frentier-zone 26
2.2. Linear barriers and frontier-zones 30
2·3· The frentier at the end of the first century AD 42
2-4- The frentier in the time of Hadrian 46
2·5· The frentier in the Severan age 50
2.6. Linear barriers and transhumant routes in the
fron tier-zone 64
4·1. Dated epitaphs of the frentier-zone 108
4. 2. The epigraphic curve at Theueste 110
4·3· Epitaphs at Theueste dated to 25-year periods I I I
4·4· Find-spots of marriage-epitaphs in the fron tier-zone I I8
4·5· Intermarriage in the frentier-zone 121
4. 6. Intermarriage at Lambaesis 12 4
4·7· Intermarriage at Theueste 126
4. 8. Intermarriage at Auzia 128
4·9· Intermarriage at Thubursicu Numidarum 13 0